Transformation of industrial planning in Singapore: Study on the

microclimatic condition of different industrial estates

Steve Kardinal Jusuf
a, *
, Nyuk Hien Wong
b
, Zhi Ying Wong
b
, Erna Tan
b
a
NUS Environmental Research Institute, National University of Singapore, 4 Architecture Drive, Singapore 117566, Singapore
b
Department of Building, National University of Singapore, 4 Architecture Drive, Singapore 117566, Singapore
a r t i c l e i n f o
Article history:
Received 24 January 2014
Received in revised form
10 April 2014
Accepted 8 May 2014
Available online 2 June 2014
Keywords:
Industrial
Microclimatic condition
Air temperature
Urban morphology
Singapore
a b s t r a c t
Despite the progression and variation of Singapore industrial developments over time, present literature
have yet to examine the changes in the urban morphology of local industrial estates and consequently,
their impact on the microclimate.
The paper discusses the microclimatic condition and urban morphology in four different industrial
developments developed by JTC in Singapore, constructed in different decades, and categorized into
Business Park development (BPD) and B2 development (B2).
Comparisons of urban morphology and microclimatic condition among the different estates were
analysed. In general, it is found that, compared to B2, BPD has a lower mean SVF value, more extensive
greenery, higher building height and building mass, a signicantly larger proportion of greenery but
lesser pavements and building area, and lower building density. There is a decreasing trend in the mean
ambient air temperature observed along with the transformation of industrial developments over time.
B2 has higher mean air temperature than BPD at all times of the day. B2 has lower diurnal temperature
range than BPD which infers that BPD is generally more effective in reducing the heat island intensity.
Finally, critical morphology parameters on ambient temperatures in industrial estates were identied.
Within the 50 m radius, critical parameters with signicant inuence on the minimum temperature
(T
min
) and the average temperature (T
avg
) values are the green plot ratio (GnPR), the total tree leaf area
(TREE) and the percentage of greenery area (GREEN), whereas the critical parameters with signicant
inuence on the maximum temperature (T
max
) are SVF, GnPR, TREE and GREEN.
2014 Elsevier Ltd. All rights reserved.
1. Introduction
The management of urban growth is a key to the spatial dis-
tribution and variability, which in turn affects the environmental
quality. Differences in land-use, surface characteristics, urban
morphology and atmospheric conditions give rise to variations in
the thermal environment within an urban area [2e7]. The impact of
land use including commercial, residential, agricultural, industrial
uses has been extensively researched using remote-sensing tech-
nology in United States, China and Germany [8e11]. More impor-
tantly, several overseas studies and Singapore have found that
regions associated with industrial activities tend to exhibit higher
temperatures than the surroundings [12].
Yu [13] noted that Singapore economy has witnessed a rapid
expansion of industrial developments since independence in 1965.
According to Han [14], approximately 7000 ha of industrial land
and 4 million m
2
of ready built factor space have been developed in
the last 36 years. The industrial sector is found to be the largest
consumer of electricity constituting 40.2% of the nationwide elec-
tricity consumption in 2011 [15]. Currently, the industrial sector
takes up 18% of the total land mass and is projected to increase to
20% by 2050 [16]. With the overall supply of industrial spaces set to
grow, these trends underline the value of examining the trans-
formation of existing industrial estates to help negate the adverse
microclimatic effects for future developments.
Despite the variations in forms and functionalities, Peddle [17]
denes industrial estates as a large tract of land, subdivided and
developed for use of several rms, distinguished by sharable
infrastructure and close proximity, bounded by the restrictions on
the lot sizes, access and utilities.
Established in 1968, the Jurong Town Corporation (JTC) is a
statutory board in charge of the planning, development, leasing and
* Corresponding author. Tel.: 65 65164691.
E-mail addresses: stevekj@nus.edu.sg, stevekardinaljusuf@gmail.com
(S.K. Jusuf).
Contents lists available at ScienceDirect
Building and Environment
j ournal homepage: www. el sevi er. com/ l ocat e/ bui l denv
http://dx.doi.org/10.1016/j.buildenv.2014.05.015
0360-1323/ 2014 Elsevier Ltd. All rights reserved.
Building and Environment 80 (2014) 48e60
management of Singapore industrial estates. With control over 80%
of the total industrial land, JTC is the largest industrial landlord and
is responsible for the implementation of industrial land use policies
to meet the changing and diverse needs of businesses [18].
Faced with the constraints of limited land, strong physical
planning framework and administration are essential to optimize
land use in view of various competing uses [13]. To prevent urban
sprawl and achieve an orderly distributed land use, boundaries,
zoning regulations and maximum plot ratio were set out in the
Master Plan. Although the Master Plan has undergone several re-
visions since its inception in 1960s, the emphasis on environmental
management is evident in the detailed planning of industrial areas.
Heavy industrial areas are concentrated southwest to the central
business district and the remaining areas are located along borders
of island, away fromthe cleaner commercial and residential estates
[19].
The emergence of different industrial developments since the
inception of Master Plan in 1960s can be categorized into three
different time frames.
During the period of 1960se1970s, ready-built standard fac-
tories and light industrial estates were constructed to facilitate the
efcient set up of labour-intensive industries. These were typically
located in suburban areas and near high density housing estates.
Flatted factories, small workshop complexes and landed factories
were the predominant forms of industrial infrastructure [20].
During the period of 1970se1980s, industrial restructuring was
necessary to retain global economic competitiveness. The switch to
capital and technology intensive industries promoted the growth of
high-value industrial businesses, thus the emergence of Science
Parks and Business Parks. The high land cost and land scarcity
necessitated land intensication. With reclamation, en-bloc rede-
velopment and higher plot ratios set out in JTC's Industrial Land
Plan for 21st Century (IP21), multi-storey premises such as stack up
and ramp up factories emerged, raising the plot ratio from 0.5 to
2.5. Specialized clusters also housed and promote knowledge-
based activities [19].
During the period of 2000se2010s, an integrated planning
approach was adopted to realize the concept of live-work-play-
learn hub, in pursuit of holistic development and sustainability of
industrial developments, with an emphasis on environmental
quality and aesthetics [20].
Under the Master Plan 2008, the industrial land use is catego-
rized into different zones according to the potential environmental
pollution. Lands zoned as Business 1 (B1) and Business 2 (B2) are
industrial developments with at least 60% of gross oor area used
for light manufacturing or warehousing purposes, leaving a
maximum of 40% for ancillary activities such as ofces and show-
rooms [14]. Another category is the Business Parks, set aside for
non-pollutive industries and businesses with at least 85% of the
total oor area used for its operations. This predominantly includes
high-technology, value-added and knowledge-intensive activities
[21].
The aim of this study is to evaluate the microclimate with
respect to the urban morphological characteristics for the different
forms of industrial estates developed by JTC in Singapore.
2. Methodology
2.1. Object of study
The paper is conned to the study of microclimatic condition
and urban morphology changes in four different industrial de-
velopments constructed in different decades. These include an in-
dustrial estate with single-user landed factories, and one estate
comprising of a mix of landed and ready-built multiple-users
atted factories, which both are categorized as B2 development.
Two different forms of business parks categorized as Business Park
Development (BPD), i.e. a pioneer development and an integrated
industrial hub are included. They are located at Loyang Industrial
Estate (Loyang), Woodlands Industrial Park (Woodlands), Interna-
tional Business Park (IBP) and one-north. The general site locations
are highlighted in Fig. 1. All four sites chosen are owned by JTC to
minimize hidden factors and allowa common basis for comparison
to study the transformation of industrial infrastructure over time.
Background information for the various sites has been obtained
from JTC.
Loyang Industrial Estate was developed in 1974e1977. Approx-
imate site study gross oor area (GFA) is 1.16sqkm with gross plot
ratio of 2.5. It is located on the eastern of Singapore. This devel-
opment is categorized as B2 development under Master Plan
(2008). Single-storey landed factories were built for aviation in-
dustrial centre, manufacturing, repair, overhaul, aviation works and
precision parts manufacturing.
Woodlands Industrial Park was developed in 1995e1999.
Approximate site study GFA is 1.78sqkmwith gross plot ratio of 2.5.
It is located on the northern of Singapore. This development is
categorized as B2 development under Master Plan (2008). Proto-
type factories (standalone/semi-detached factories) were built for
food production and manufacturing from raw materials, assembly
of plants and machinery, and treatment of waste oil/solvents.
IBP was developed in 1992e2009. Approximate site study GFA is
0.37sqkmwith gross plot ratio of 1.0e2.5. It is located on the south-
Nomenclature
JTC Jurong Town Council
BPD Business Park development
B2 B2 development
SVF sky view factor
T
min
minimum temperature
T
avg
average temperature
T
max
maximum temperature
GnPR Green Plot Ratio within 50 m radius area
TREE total tree leaf area within 50 m radius area
GREEN percentage of greenery area within 50 m radius area
IP21 Industrial Land Plan for 21st Century
LTA Land Transport Authority
GIS Geographic Information System
BDG percentage of building footprint area within 50 m
radius area
PAVE percentage of pavement area within 50 m radius area
HBDG average height to building area ratio
WALL total wall surface area
AvgHT average building height
TURF turf area within 50 m radius area
TotalGreen total green area within 50 m radius area
DTR diurnal temperature range
Gross oor area all covered oor areas of a building, except
otherwise exempted, and uncovered areas for
commercial uses [1]
Gross plot ratio the ratio of the gross oor area of a building(s) to
its site area [1]
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 49
western of Singapore. This development is categorized as Business
Park Development (BPD) under Master Plan (2008). Purpose-built
high rise buildings were built for information technology (IT),
software development, research and development (R&D), high-
tech manufacturing, laboratory testing and data processing
activities.
One-north is developed in 2005e2025. Approximate site study
GFA is 2.36sqkmwith gross plot ratio of 3.3e7.5. It is located on the
south-western of Singapore. This development is categorized as
BPD under Master Plan (2008). Mixed-use development (high rise)
are built for educational institutions, residences and recreational
amenities, research facilities and ofces.
Several measurement points were chosen to cover different
urban morphology in each area. In Loyang, there were 11 mea-
surement points. In Woodlands, there were 12 points. In IBP, there
were 10 points. In one-north, there were 30 points. Locations of
measurement points can be referred to in Fig. 1.
2.2. Climate predictor parameters
On each measurement point, U12-011 HOBO temperature/RH
data logger and solar cover were used to measure the ambient air
temperature. Wind speed smart sensor S-WSA-M003 and HOBO
micro station H21-002 were used to measure the wind speed
condition. All sensors were installed on lamp posts at 2.1 m
above ground in compliance to Land Transport Authority (LTA)
regulations. For Loyang and Woodlands, data logging was
congured at 1-min interval, while IBP and one-north were
congured at 10-min and 5-min interval respectively, which
were all averaged into hourly means. The sensors were factory
calibrated.
As the reference point, each area had one weather station
installed as shown in Fig. 1 for Loyang, Woodlands and IBP. Analysis
for one-north used meteorological data from nearby weather sta-
tion in National University of Singapore (managed by Department
of Geography) [23]. HOBO Weather Station H21-001 was used to
collect data of solar radiation (W/m
2
), ambient air temperature (

C),
relative humidity (%), wind direction (degree), wind speed (m/s)
and rain fall (mm).
2.3. Urban morphology parameters
To obtain the urban morphology data, master plan drawings of
the four industrial estates were input into the Geographic Infor-
mation System (GIS) system to extract the relevant morphology
information. In quantifying the morphology parameters, a buffer of
50 m radius with respect to each the measurement point was
chosen. Kruger and Givoni [24] examined three radii, 56 m, 125 m
and 565 m, and found 56 m to be most effective in producing sig-
nicant correlation between land use patterns and local tempera-
tures. Similarly, in a local study, Jusuf and Wong [25] concluded that
50 m radius had the most signicant inuence amongst the 4 radii
(25 m, 50 m, 75 m and 100 m) based on the correlation strength
between building area and T
min
, T
avg
and T
max
.
Using ArcGIS, the two dimensional area of greenery (GREEN),
building (BDG) and pavement (PAVE) within the radius is quanti-
ed. The quantity of trees on site within the bound area is veried
by site survey. The total leaf area is then obtained according to the
weightage assigned as set out in the equation developed by Ong
[26] for the calculation of green plot ratio (GnPR).
The average height to building area ratio (HBDG) and total wall
surface area (WALL) can also be calculated based on the height
attribute data in BDG extracted from the GIS measurement. Ac-
cording to Giridharan et al. [27], the thermal mass of built envi-
ronment within the area of inuence radius is a function of HBDG,
derived by dividing the average height of buildings over the oor
area.
Similar to the methodology detailed by Svensson and Eliasson
[28], the photographical method involving sh-eye lens and Nikon
Digital camera is used to capture the sky view factor (SVF) for each
measurement point. The images are then imported into the Ray-
Man 1.2 software and converted into black-and-white form, with
the sky in white and buildings or trees are black for SVF calculation
[29].
Fig. 1. Locations of Loyang, Woodlands, IBP and one-north Industrial Estates mapped onto URA Master Plan 2008 [22] and the corresponding measurement points.
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 50
Nikon D80 Digital camera with Nikon AF DX 10.5 F2.8G IF ED
sheye lens was used to capture the SVF.
2.4. Data selection for analysis
The eld measurements were conducted in different time
frames. They were 1 July 2012e31 December 2012 for Loyang and
Woodlands; 1 December 2010e30 April 2011 for IBP; and 1
November 2007e31 March 2008 for one-north.
Using the meteorological data, the data analysis focuses on
fairly clear, calm (wind speed less than 3 m/s) and hot weather
condition, selected by analysing the prole of daily solar radia-
tion which exhibits a bell curved shape with a peak exceeding
700 W/m
2
, wind speed less than 3 m/s, air temperature and no
rain.
Based on the selection criteria, dates were selected from the
respective measurement periods for which the corresponding air
temperatures were further analysed. Loyang used total of 29 days
data which 11 days were from July 2012, 4 days were from August
2012, 3 days were from September 2012, 8 days were from October
2012, 2 days were from November 2012 and 1 day was from
December 2012. Woodlands uses total of 23 days data which 7 days
were from July 2012, 6 days were from August 2012, 3 days were
from September 2012, 5 days were from October 2012, 1 day was
from November 2012 and 1 day was from December 2012. IBP uses
total of 23 days data which 2 days were from December 2010, 6
days were from January 2011, 8 days were from February 2011, 3
days were from March 2011, and 4 days were from April 2011. One-
north uses total of 36 days data which 6 days were from November
2007, 7 days were from December 2007, 6 days were from January
2008, 15 days were from February 2008, and 2 days were from
March 2008.
2.5. Data analysis
From the eld measurement data, the urban morphology pa-
rameters were compared for the industrial estates. The urban
morphology parameters discussed were SVF, GnPR, average
building height (AvgHT) and total wall surface area (WALL),
average building height to oor area ratio (HBDG) and composi-
tion of pavement (PAVE), greenery (GREEN) and building footprint
(BDG).
Fromthe urban morphology analysis, comparison of ambient air
temperature among estates and comparison of average ambient air
Fig. 2. SVF measurement results.
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 51
temperatures between categories of industrial were made. The
distribution of ambient air temperature was analysed. Daytime
refers to 08:00 to 19:00 h, whereas nighttime refers to 20:00 to
07:00 h.
KruskaleWallis test was used to see if there is any statistical
difference for each parameter among the four industrial estates,
while ManneWhitney U test was used to nd signicant difference
for each parameter between the two categories of industrial
development. Post-hoc pair-wise test was further used to reveal
any difference for each parameter between two of the industrial
estates.
Finally, critical morphology parameters affecting the ambient
temperatures in industrial estates were analysed. Pearson correla-
tion was used to measure the linear correlation between air tem-
perature and each parameter.
3. Urban morphology of industrial estates
3.1. Comparison of SVF measurement results
Fig. 2 shows the sky view factor (SVF) measurement results
obtained.
As shown in Fig. 3a, Loyang exhibits a moderately high SVF
range from 0.65 to 0.85, which concurs with the fact that it is a
relatively open area with short buildings located apart. The pres-
ence of foliage is consistently observed across all the measured
spots, which indicates well spread greenery despite the generally
sparse canopy. Shading is provided mainly by trees from the
immediate surroundings, with minimal sky obstruction by build-
ings; except Point 7 where the lowest SVF of 0.37 is observed due to
the extensive crown of a single tree with dense canopy with mul-
tiple overlapping leaves.
As shown in Fig. 2b, Woodlands exhibits either a high SVF of
above 0.90 or below that of 0.70. The observed values indicate an
uneven distribution of greenery across the various spots in the
industrial park. In Points 3, 6, 7, 8 and 10, majority of the shading is
caused by dense tree canopies with some encroachment of sur-
rounding buildings into the sky visibility as shown from the
completely black portions of SVF images. As for Points 1, 2, 4, 5, 9, 11
and 12, the obvious lack of adjacent building structures and sparse
leafage barely provides any effective shade, thus contributing to
high SVF value.
As shown in Fig. 2c, IBP displays a narrower range of SVF values
between 0.65 and 0.85 as compared to Woodlands. The effective-
ness of shade is relatively consistent across all the spots. While
similar values are found in Loyang, the distinct difference is that
instead of well-spread sparse foliage, the dark parts here are
completely shaded and highly concentrated around the perimeters
of the hemisphere captured. A large part of shading in IBP is due to
the buildings from the immediate surroundings rather than the
tree canopy, as inferred from the evident polygonal shaded outline.
This implies that the building masses are generally taller and more
closely arranged in the estate layout.
As shown in Fig. 2d, a wide range of SVF values from0.15 to 0.99
are observed in one-north, with an equal number of spots with
values above and below 0.50. For points where SVF value falls
Fig. 2. (continued).
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 52
below 0.50, the shaded cover is either completely due to the
buildings that ank the sides of the measured spot as in Points 3
and 4, or the canopy of mature trees as in Points 7, 17, 24 and 28. The
exceptions with a combination of both are at Points 21 and 22. A
mix of the multi-storied industrial complexes and foliage from
trees lining the streets reduce the overall sky visibility.
Table 1 show the comparison of the SVF values observed across
the four industrial estates. The variation in SVF values is the least in
IBP with a standard deviation of 0.072, whereas the values vary the
most in one-north with 0.302.
Comparing the mean value of the SVF results, the largest dif-
ference of 0.172 is again noted between IBP and one-north, thus,
although both are business park developments, the former is
generally less shaded throughout the site.
Although the SVF values vary quantitatively, the KruskaleWallis
test results show insufcient evidence to conclude any statistical
difference among the four industrial estates.
The ManneWhitney U test results reveal no signicant differ-
ence between the SVF values for B2 and BPD. Nonetheless, BPD has
a lower mean SVF value (0.622) than B2 (0.725). This implies that
on average, any spot within BPD receives a greater extent of shade
than that of B2.
3.2. Comparison of greenery condition
In the evaluation of greenery condition for the four industrial
estates, the total tree leaf area (TREE) and turf area (TURF) are the
two key components used to quantify the Green Plot Ratio (GnPR).
The GnPR is derived from Equations (1) and (2).
Total Tree Leaf Area No: of Trees Canopy Area LAI
(1)
GnPR
Total Tree Leaf Area Turf Area
Area of Circler 50 m
(2)
Table 2 provides the descriptive statistics summarizing the
comparison of the three parameters. Total green area (TotalGreen)
is the total of TREE and TURF within an area of 50 m radius.
As shown in Table 2, one-north has the largest total green area
for an average spot, followed by IBP, Woodlands and Loyang with
the least. Interestingly, the composition that adds up to the total
greenery for each estate is different.
Although one-north does not have the most extensive turf area,
it possesses the largest total tree leaf area, which hence attains the
highest total greenery area. It can be inferred that the main con-
tributors of foliage are trees, possibly due to the presence of
forested parks within the estate. The high density of leaves is in
agreement with the low SVF values detailed earlier.
Conversely, for IBP, the predominant greenery on site is turf and
its extensiveness compensates for the small tree leaf area to bring
the total greenery to the second highest. Similar to one-north,
Woodlands has limited turf area, thus the total tree leaf area
plays the key role in contributing to the overall green condition.
Loyang has equally low proportion of turf and tree leafage, thus
comparatively it has the poorest greenery condition among the
four. Comparing Woodlands and Loyang, the former has an
evidently higher tree leaf area yet smaller turf, which implies the
Fig. 2. (continued).
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 53
Fig. 2. (continued).
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 54
on-site presence of more trees with extensive leafage than the
latter.
As expected, since it corresponds with the total greenery, the
highest average GnPR is observed in one-north, followed by IBP,
Woodlands and Loyang. Comparing the two estates with the
greatest difference for average GnPR, one-north has 25.17% more
greenery than Loyang for every square metre on site.
Both the highest and lowest GnPR values are observed in one-
north; with GnPR of 3.73 measured at Point 29 located within a
park where dense clusters of trees are found, whereas the GnPR of
0.26 is found at Point 4 within a high-rise built-up area. Such low
values are the result of the high concentration of built structures
and larger building footprint, which proportionately reduces the
turf area.
Based on the KruskaleWallis test results, there is a statistically
signicant difference in the GnPR values across the four industrial
estates. The post-hoc pair-wise test further revealed a real differ-
ence in the greenery condition between Loyang and one-north, and
Loyang and IBP.
The ManneWhitney U test results verify that the GnPR of two
types of development are signicantly different. As inferred from
the higher average rank of BPD than B2, it can be concluded that in
general BPD has more extensive greenery than B2 development.
3.3. Comparison of average building height
Among the four estates, IBP has the highest average building
height of 20.13 m, which is 2.41 times taller than that of Loyang. As
detailed in Table 3, compared to Woodlands and Loyang, one-north
and IBP have a much higher standard deviation, which translate to
a wider range of building heights observed.
Hence it can be inferred that Loyang is characterized by build-
ings of generally similar heights between 6 m and 10 m, whereas
for Woodlands, IBP and one-north, some buildings are twice the
height of others, which verify the different types of facilities found
on site, such as multi-storeyed ready-built factory and single-storey
landed factories.
The KruskaleWallis test results conrm that there is a signi-
cant difference in average building height (AvgHT) in the four es-
tates, but the post-hoc pair-wise test results show insufcient
evidence to distinguish the statistically different pairs of industrial
estate.
Nonetheless, the quantitative difference in the average building
height within the same development category is minimal. The
disparity between Loyang and Woodlands, as well as one-north and
IBP is only 1.69 m and 1.11 m respectively. This is relatively insig-
nicant considering that a typical oor-to-oor height for an in-
dustrial building is 3.8 me5.0 m [30].
Comparing the average building height of BPD and B2 devel-
opment, a substantial difference of 52.07% is noted. The Man-
neWhitney U test results conrm that the BPD has a signicantly
higher median height than B2 developments.
Interestingly, there appears to be a positive relationship be-
tween maximum height of buildings and the transformation of
industrial developments over time, with increasingly taller build-
ings are developed in the recent years. A driving factor could be the
need to maximize industrial land use, through the allocation of
higher Gross Plot Ratio achievable under the planning policy. Also,
along with the shift in the economic focus to high value industrial
businesses and operations, the change in function also necessitates
a change in the building form.
3.4. Comparison of total wall surface area
The total wall surface area (WALL) increases when the buildings
get higher, as it is computed as a function of the building height and
Table 2
Comparison of greenery condition among the four industrial estates.
N Minimum Maximum Mean Std.
deviation
Loyang Turf 11 2791.72 5524.73 4220.29 823.40
Trees 11 2199.96 17,220.98 4647.82 4310.30
TotalGreen 11 5901.33 22,745.71 8868.11 4755.89
GnPR 11 0.75 2.89 1.13 0.61
Woodlands Turf 12 1278.76 4620.55 3063.19 869.43
Trees 12 2829.75 15,470.00 6335.58 4158.28
TotalGreen 12 4108.51 20,090.55 9398.78 4837.37
GnPR 12 0.52 2.56 1.20 0.62
IBP Turf 10 4737.68 6375.44 5638.67 586.30
Trees 10 1611.32 6701.98 4736.95 1887.50
TotalGreen 10 7876.00 12,037.12 10,375.62 1774.64
GnPR 10 1.00 1.53 1.32 0.22
One-north Turf 29 470.84 6867.29 4766.95 1841.44
Trees 29 0.00 22,476.95 7059.01 5768.42
TotalGreen 29 2070.17 29,268.76 11,825.96 7085.15
GnPR 29 0.26 3.73 1.51 0.90
Table 3
Comparison of average building height among the four industrial estates.
N Minimum Maximum Mean Std. deviation
Loyang 11 7 15.75 8.37 2.76
Woodlands 12 0 31.5 10.06 9.79
IBP 10 0 45.5 20.13 14.08
One-north 29 0 108 19.02 19.94
The value 0 indicates absence of buildings.
Fig. 3. Comparison of building, pavement and greenery percentage composition.
Table 1
Comparison of SVF results among the four industrial estates.
Location Mean N Std. deviation Minimum Maximum Range
Loyang 0.722 11 0.140 0.37 0.9 0.53
Woodlands 0.728 12 0.238 0.32 0.96 0.64
IBP 0.751 10 0.072 0.66 0.89 0.23
One-north 0.579 29 0.302 0.00 0.99 0.99
Total 0.660 62 0.250 0.00 0.99 0.99
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 55
length within the radius of inuence. It does not take into account
allowances or deduction for openings or extensions here.
As shown in Table 4, one-north has the largest average wall
surface area, followed by Woodlands, Loyang andIBP. Althoughone-
north has a taller average building height than Woodlands, the
WALL values are comparable with a small difference of only 6.37%.
This implies that the large wall surface area is mainly due to its tall
building height. In contrast, the morphology in Woodlands is char-
acterized by shorter but more number of buildings, which add on to
the building lengths on the two-dimensional foot print or building
perimeter within radius of inuence. Despite the quantitative dif-
ference, the average WALL values across the four estates were not
signicantly different based on the KruskaleWallis test results.
Comparing average WALL values, BPD has a slightly larger
average total wall surface area at 2718.52 m
2
than that of B2 at
2405.97 m
2
by 11.50%, due to generally taller buildings more closely
located to each other. This results in higher building density.
The ManneWhitney U test results reveal no signicant difference
inthe average WALL values betweenthetwocategories. However, the
maximum WALL value found in one-north is 3.08 times that of IBP,
whereas Woodlands is only 1.35 times that of Loyang. The variations
in maximum WALL values observed within the BPD category is thus
much larger than within the B2 developments.
Interestingly, Loyang has a higher maximum WALL value at
5107.96 m
2
than that of IBP at 3943.71 m
2
, but a corresponding
building height half that of IBP at 9.8 m compared to 21 m. This
indicates that the two dimensional footprint occupied by the
buildings are bigger, which implies higher building density than
the latter at the corresponding points measured. One-north shows
the highest maximum WALL value and height of all at 12,136.58 m
2
and 38 m respectively. The highest maximum WALL value and
height in Woodlands are 6913.62 m
2
and 31.5 m respectively.
3.5. Comparison of average building height to building area ratio
The average height to building area ratio (HBDG) is an important
indication of the thermal mass in the estate. This has implications
on the ambient air temperatures [27].
IBP has the highest value, followed by one-north, Loyang and
Woodlands. As deduced from Table 5, on average for every square
metre, the building height in IBP is 1.66, 6.57 and 12.28 times that of
one-north, Loyang and Woodlands respectively.
Statistically, a distinct difference in the HBDG is observed from
the KruskaleWallis test results. The paired-wise test further shows
that the signicant differences are located between the pairs of
Loyang and Woodlands versus one-north. Through the Man-
neWhitney U test, a real difference is noted between the B2 and
BPD HBDG values. This is in line with the visual observations from
the site surveys where BPD have signicantly higher building mass
than that of B2 developments.
3.6. Comparison of composition of building, pavement and greenery
From Fig. 3, comparing the building footprint area (BDG),
Woodlands appears to have the highest value percentages, fol-
lowed by Loyang, one-north and IBP. A similar order is observed in
the case of pavement (PAVE) percentages. However, for greenery
(GREEN) percentage, a reverse trend is noted, with the highest
values noted in IBP, one-north then Loyang and Woodlands.
The KruskaleWallis test results prove that there is a signicant
difference among the percentage compositions for all building,
pavement and greenery. From the post-hoc tests, it is inferred that
on average compared to Woodlands, one-north has a signicantly
smaller proportion of the estate occupied by buildings in terms of
two-dimensional footprint by 9.41%. Woodlands also has a signi-
cantly larger proportion of estate occupied by roads and sidewalks
than that of IBP by 23.23%.
However, IBP and one-north have signicantly greater propor-
tion of their sites occupied by turf, such as grass, by a difference of
32.76% and 21.67% respectively than Woodlands.
Summarizing the ManneWhitney U test results, a statically
signicant difference in the composition between the B2 and BPD
can be concluded. Generally, BPD sites comprise of a signicantly
larger proportion of greenery but lesser pavements and building
area than B2 developments.
FromTable 6, the morphology parameters that differ statistically
across the four estates and B2 and BPD are the GnPR, the average
building height, the average height to building area ratio, and
percentage composition of building, pavement and greenery. Thus,
it can be inferred that there is a distinct change in the urban
morphology of industrial estates with its transformation over time.
4. Comparison of microclimatic condition within industrial
estates
4.1. Comparison of ambient air temperature results among estates
The diurnal temperature prole for each estate was plot using
the mean hourly temperatures, by averaging the hourly readings
fromall measurement points for the selected days. Fig. 4 shows the
comparison of the average air temperatures among four industrial
estates.
It is observed that Loyang generally exhibits the highest average
temperatures, followed by Woodlands, IBP and one-north. The
maximum temperatures during daytime occur at around 14:00 to
15:00 h. The highest average temperature of 32.70

C at Loyang is
0.13

C slightly higher than that of Woodlands and 1.68

C higher
Table 4
Comparison of average wall surface area among the four estates.
N Minimum Maximum Mean Std. deviation
Loyang 11 384.09 5107.96 1996.10 1318.59
Woodlands 12 0 6913.62 2781.69 2134.68
IBP 10 0 3943.71 1986.39 1443.48
One-north 29 0 12,136.58 2970.97 2573.22
Table 5
Comparison of HBDG values among the four estates.
N Minimum Maximum Mean Std. deviation
Loyang HBDG 11 0.26 3.72 0.859 0.992
Woodlands HBDG 12 0 1.45 0.456 0.434
IBP HBDG 10 0 42 5.649 12.869
One-north HBDG 29 0 23.34 3.297 4.902
Table 6
Summary of statistical conclusions for morphology parameters compared.
Morphology parameters Statistically different at a 0.05 level
Loyang, Woodlands IBP
and one-north
B2 and BPD
SVF No No
GnPR Yes Yes
Average Building Height Yes
a
Yes
Average Total Wall Surface area No No
Average Height to Building Area Ratio Yes Yes
Building Area Percentage Yes Yes
Pavement Percentage Yes Yes
Greenery Percentage Yes Yes
a
But insufcient to distinguish signicantly different pairs.
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 56
than that of IBP. The largest difference of 1.87

C is noted when
compared to one-north.
During nighttime, the lowest air temperatures are observed at
06:00 to 07:00 h. Among the four estates, one-north is the coolest
estate with the lowest average temperature recorded at 24.39

C.
This is 0.35

C, 1.89

C and 1.99

C less than that of IBP, Woodlands
and Loyang respectively.
IBP has the highest diurnal temperature range value of 6.49

C
(DTR maximum minimum temperatures) whereas Woodlands
has the smallest DTR of 6.22

C. DTR in Loyang and one-north are
6.32

C and 6.36

C respectively. Such thermal behaviour is mainly
inuenced by the surrounding morphology properties. According
to Gallo and Owen [31], a lower DTR is an indication of extensive
urbanization change in land cover. Earlier in Fig. 3, Woodlands has
the highest percentage of estate occupied by buildings and pave-
ment. The extensive concrete and impervious surfaces result in
greater absorption of solar radiation, while proportionally reducing
the green surfaces for evapotranspiration, thus contributing to
higher air temperatures.
The smallest and largest temperature differences among the
estates are noted between Loyang and one-north, at 13:00 h and
16:00 h respectively. Interestingly, this occurs right before and after
the period where the maximum temperatures are observed. Since
one-north has a much higher mean GnPR value than Loyang, the
presence of extensive foliage helps to effectively lower the ambient
air temperature via the evaporative cooling of surface moisture,
thus attaining a lower and more gradual peak compared to Loyang.
As shown in Fig. 4, one-north typically has the lowest mean
temperature throughout the day, except from 08:00 to 13:00 h,
where temperatures are 0.49

Ce0.93

C higher than IBP. Although
it receives extensive shading based on the low SVF values, the
cooling effects seem to be outweighed by heat gain from other
sources. This is possible since the differences among the SVF values
are not statistically signicant.
A KruskaleWallis test is used to determine if there is a group
difference among the average ambient air temperatures found in
the four industrial estates. The null hypothesis is that the distri-
bution of average ambient air temperature values is the same for
Loyang, Woodlands, IBP and one-north.
The results in Table 7 suggest that at 95% condence level, there
is a statistically signicant difference among the ambient air tem-
perature distribution for the four estates, with H(3) 323.93,
p < .001. Since p < .05, the null hypothesis is rejected.
From the pair-wise test, the temperature values for IBP and one-
north versus Woodlands and Loyang are statistically signicantly
different, since p-values are less than 0.05. However, there is no
distinct difference in the air temperatures within the same category
of development.
Generally, a decreasing trend in the mean ambient air temper-
atures is observed along with the transformation of industrial de-
velopments over time, with temperature differences most distinct
between the categories of development.
4.2. Comparison of average ambient air temperatures between
categories
Fig. 5 shows that B2 developments have higher mean air
temperature than BPD at all times of the day. On average, the
temperature difference is of 1.71

C. The average maximum
temperatures of 32.59

C and 30.99

C are observed at 15:00 and
14:00 h respectively. The lowest average temperatures of
26.33

C and 24.59

C occur at 06:00 h in the early morning
before sun-rise.
The largest temperature difference between the two estates of
2.14

C was observed at 17:00 h, while the smallest difference of
1.16

C occurred at 12:00 noon. For BPD, the lower temperatures are

primarily due to the shading effects of the trees and tall buildings,
which help moderate absorption of solar radiation on the sur-
rounding surfaces. However, at noon, the position of sun is directly
above the buildings, thus the cooling effect of shadows from
buildings is considerably minimized. Together with the maximum
solar penetration, the heat gain outweighs the cooling effects, thus
resulting in higher temperatures and a reduced difference relative
to B2.
Fig. 4. Comparison of diurnal air temperature distribution for Loyang, Woodlands, IBP
and one-north.
Table 7
KruskaleWallis test on average ambient air temperature values.
Location N Mean rank
T
avg
Loyang 696 1637.95
Woodlands 552 1591.06
IBP 552 1108.25
One-north 864 1064.52
Total 2664
T
avg
Chi-Square 323.927
df 3
Asymp. Sig. 0.000
a. Kruskal Wallis Test
b. Grouping Variable: Location
Fig. 5. Comparison of average ambient air temperatures between B2 and BPD.
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 57
A ManneWhitney U test is used to examine if the average
temperatures differ statistically between the two categories of
development. The null hypotheses are that the average tempera-
ture values for B2 are equal to BPD. The results of the statistical test
are shown in Table 8.
Since the ManneWhitney U-test results are signicant, p < .05,
thus, the null hypothesis can be rejected with 95% condence.
There is sufcient evidence to conclude that there is statistical
difference in the average temperature values between B2 and BPD.
B2 has a lower diurnal temperature range at 6.26

C than BPD at
6.40

C, which is a result of warmer minimum air temperatures
relative to the maximumtemperatures. The minimumemaximum
temperatures for B2 and BPD are 26.33

Ce32.59

C and
24.59

Ce30.99

C respectively. It can be inferred that BPD is
generally more effective in reducing the heat island intensity. Such
difference in the thermal performance has practical implications on
the energy usage. For every 1

C lowered, the energy consumption
due to cooling load can be reduced by 5%, which translates into
substantial monetary and energy savings in the long run [32].
5. Critical morphology parameters on ambient temperatures
in industrial estates
To determine the most important morphology parameters that
affect the ambient air temperature in industrial estates, a multiple
regression analysis is performed. From the correlation coefcient,
the signicance of each parameter on the air temperature can be
quantied.
In Table 9, the regression results for minimum temperature
(T
min
) and average temperature (T
avg
) models both show a strong
correlation coefcient of 0.909 and 0.799, whereas maximum
temperature (T
max
) has a moderate correlation of 0.69 respectively.
All the regression models have a signicance of 0.00. The TURF and
PAVE variables have been excluded in the regression models for all
T
min
, T
avg
and T
max
as they have a collinearity of near zero with other
predictors, which would otherwise greatly reduce the statistical
power of the model.
The regression ndings on morphology predictors and air
temperature can be summarized as follows:
The GnPR variable shows a negative sign in both the regression
coefcient and Pearson correlation. This is in agreement with
the general theory that a high GnPR would lead to lower
ambient air temperatures.
The SVF variable shows a positive sign for the regression coef-
cient and Pearson correlation values in T
avg
and T
max
models.
This is in line with the general theory that higher SVF values
result in higher ambient temperatures since there is reduced
obstruction, allowing the solar radiation to penetrate and heat
up the surfaces [33,34]. However, a negative regression coef-
cient is observed for T
min
model. The minimum temperatures
are typically observed during nighttime in the absence of strong
solar radiation, thus the shading effects by buildings or foliage
have little cooling effect compared to daytime.
The HBDG variable has different signs in the regression coef-
cient and Pearson correlation, showing that the collected data in
this industrial development study is not sufcient enough to
show conclusive correlation between HBDG variable and
ambient air temperature.
The WALL shows negative correlation with T
avg
and T
max
,
although data is not sufcient enough to show conclusive cor-
relation with T
min
. This phenomenon explains that larger WALL
during daytime provides shading and therefore, reduces
ambient air temperature (T
avg
and T
max
). Larger WALL during
daytime absorbs heat and releases it at night, resulting in higher
ambient air temperature (T
min
), which is found in our other
study [25] showing positive correlation.
As inferred from morphology variables with 2-tailed signicant
values of p < .05, the critical parameters with signicant inuence
on the T
min
and T
avg
values are GnPR, TREE and GREEN, whereas the
critical parameters with signicant inuence on the T
max
are SVF,
GnPR, TREE and GREEN. The regression models verify that vege-
tation generally plays an important role to provide positive envi-
ronmental impact.
Table 8
ManneWhitney test on average temperatures for B2 and BPD.
Development N Mean rank Sum of ranks
T
avg
B2 24 29.92 718.00
BPD 24 19.08 458.00
Total 48
T
avg
ManneWhitney U 158.000
Wilcoxon W 458.000
Z 2.681
Asymp. Sig. (2-tailed) 0.007
a. Grouping Variable: Development
Table 9
Regression results of morphology parameters and air temperature.
Regression Variables SVF, GnPR, TURF, TREES, AvgHT, WALL, HBDG, PAVE, BDG, GREEN
Dependent Variables Average, Minimum and Maximum ambient air temperatures
Variables T
min
T
avg
T
max
B Sig. Pearson correlation B Sig. Pearson correlation B Sig. Pearson correlation
Constant 24.802 0.000 27.297 0.000 30.343 0.000
SVF 0.193 0.503 0.185 0.288 0.132 0.301 1.173 0.004 0.445
GnPR 3.049 0.000 0.596 3.009 0.000 0.586 2.678 0.000 0.437
TURF NA NA 0.476 NA NA 0.472 NA NA 0.278
TREE 3.58E-04 0.000 0.283 3.68E-04 0.000 0.247 3.37E-01 0.000 0.170
AVGHT 0.004 0.470 0.198 0.001 0.707 0.272 2.27E-04 0.977 0.297
WALL 6.34E-05 0.263 0.028 7.86E-05 0.037 0.073 1.25E-04 0.107 0.233
HBDG 0.001 0.890 0.23 0.003 0.704 0.225 0.005 0.733 0.148
BDG 0.019 0.076 0.427 0.021 0.054 0.381 0.031 0.035 0.203
PAVE NA NA 0.332 NA NA 0.359 NA NA 0.226
GREEN 0.026 0.000 0.76 0.025 0.000 0.472 0.028 0.000 0.277
F 30.851 76.146 17.665
R2 0.826 0.921 0.731
Adjusted R2 0.799 0.909 0.69
Std Error 0.41496 0.2722 0.56262
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 58
A correlation analysis was performed to understand if there is a
difference between the B2 and BPD in the critical morphology pa-
rameters that affect the ambient air temperature. As shown in
Table 10, the signicant variables between the two are vastly
different.
6. Conclusions
Comparisons of urban morphology and microclimatic condi-
tion among the different estates were analysed. In general, it is
found that Business Park development (BPD) has a lower mean
SVF value than B2 development (B2), BPD has more extensive
greenery in general than B2, BPD has higher building height and
building mass than B2, B2 has higher building density than BPD,
and BPD has a signicantly larger proportion of greenery but
lesser pavements and building area than B2. There is a decreasing
trend in the mean ambient air temperature observed along with
the transformation of industrial developments over time. B2 has
higher mean air temperature than BPD at all times of the day. B2
has lower diurnal temperature range than BPD which infers that
BPD is generally more effective in reducing the heat island
intensity.
During the analysis of critical morphology parameters on
ambient temperature in the industrial estates, some ndings are as
follows:
Higher GnPR leads to lower ambient air temperature;
Higher SVF values result in higher ambient temperatures for T
avg
and T
max
models due to reduced shading, allowing the solar
radiation to penetrate and heat up the surfaces;
Higher SVF values result in lower ambient temperatures for T
min
model due to the absence of strong solar radiation at night;
WALL variable reduces daytime ambient temperature (T
max
and
T
avg
) due to the shading effect.
Acknowledgement
This paper is part of the researchproject Development of Climatic
Mapping Tool for Estate Environmental Evaluation funded by Jurong
Town Corporation (JTC) Singapore grant number JTC C09402011.
References
[1] Urban Redevelopment Authority. Development control parameters for resi-
dential development [Internet] [cited 2014 Mar 31]. Available from:, http://
www.ura.gov.sg/circulars/text/dcdrhb_d0e4.htm; 2011 July.
[2] Hart MA, Sailor DJ. Quantifying the inuence of land-use and surface char-
acteristics on spatial variability in the urban heat island. Theor Appl Climatol
2009;95(3e4):397e406.
[3] Perini K, Magliocco A. Effects of vegetation, urban density, building height, and
atmospheric conditions on local temperatures and thermal comfort. Urban
Urban Green; 2014. http://dx.doi.org/10.1016/j.ufug.2014.03.003.
[4] Zhao C, Fu G, Liu X, Fu F. Urban planning indicators, morphology and climate
indicators: a case study for a north-south transect of Beijing, China. Build
Environ 2011;46:1174e83.
[5] Yang F, Lau SSY, Qian F. Urban design to lower summertime outdoor tem-
perature on high-rise housing in Shanghai. Build Environ 2011;46:769e85.
[6] Busato F, Lazzarin RM, Noro M. Three years of study of the Urban Heat Island
in Padua: experimental results. Sustain Cities Soc 2014;10:251e8.
[7] Yan H, Fan S, Guo C, Wu F, Zhang N, Dong L. Assessing the effects of landscape
design parameters on intra-urban air temperature variability: the case of
Beijing, China. Build Environ 2014;76:44e53.
[8] Lo CP, Quattrochi DA, Luvall JC. Application of high-resolution thermal
infrared remote sensing and GIS to assess the urban heat island effect. Int J
Remote Sens 1997;18(2):287e304.
Table 10
Pearson correlation results on morphology parameters and air temperature.
Correlation
B2 BPD
T
min
T
avg
T
max
T
min
T
avg
T
max
SVF Pearson Correlation 0.137 0.168 0.342 SVF Pearson Correlation 0.148 .545
b
.680
b
Sig. (2-tailed) 0.534 0.443 0.110 Sig. (2-tailed) 0.368 0.000 0.000
N 23 23 23 N 39 39 39
GnPR Pearson Correlation .449
a
.613
b
0.013 GnPR Pearson Correlation .620
b
.542
b
0.229
Sig. (2-tailed) 0.032 0.002 0.953 Sig. (2-tailed) 0.000 0.000 0.161
N 23 23 23 N 39 39 39
TURF Pearson Correlation 0.058 0.046 0.210 TURF Pearson Correlation .598
b
.392
a
0.062
Sig. (2-tailed) 0.792 0.833 0.337 Sig. (2-tailed) 0.000 0.014 0.709
N 23 23 23 N 39 39 39
TREE Pearson Correlation .448
a
.613
b
0.013 TREE Pearson Correlation .553
b
.525
b
0.294
Sig. (2-tailed) 0.032 0.002 0.954 Sig. (2-tailed) 0.000 0.001 0.069
N 23 23 23 N 39 39 39
AVGHT Pearson Correlation 0.157 0.212 0.378 AVGHT Pearson Correlation .342
a
0.187 0.033
Sig. (2-tailed) 0.474 0.331 0.075 Sig. (2-tailed) 0.033 0.253 0.842
N 23 23 23 N 39 39 39
WALL Pearson Correlation 0.163 0.061 0.286 WALL Pearson Correlation .365
a
0.026 .335
a
Sig. (2-tailed) 0.458 0.784 0.186 Sig. (2-tailed) 0.022 0.876 0.037
N 23 23 23 N 39 39 39
HBDG Pearson Correlation 0.138 .437
a
0.029 HBDG Pearson Correlation 0.027 0.152 0.167
Sig. (2-tailed) 0.531 0.037 0.895 Sig. (2-tailed) 0.873 0.361 0.317
N 23 23 23 N 38 38 38
BDG Pearson Correlation 0.212 0.413 0.025 BDG Pearson Correlation .519
b
0.127 0.202
Sig. (2-tailed) 0.332 0.050 0.909 Sig. (2-tailed) 0.001 0.443 0.217
N 23 23 23 N 39 39 39
PAVE Pearson Correlation 0.266 .440
a
0.223 PAVE Pearson Correlation .452
b
.440
b
0.053
Sig. (2-tailed) 0.220 0.036 0.306 Sig. (2-tailed) 0.004 0.005 0.747
N 23 23 23 N 39 39 39
GREEN Pearson Correlation 0.058 0.046 0.209 GREEN Pearson Correlation .598
b
.392
a
0.062
Sig. (2-tailed) 0.793 0.833 0.337 Sig. (2-tailed) 0.000 0.014 0.709
N 23 23 23 N 39 39 39
a
Correlation is signicant at the 0.05 level (2-tailed).
b
Correlation is signicant at the 0.01 level (2-tailed).
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 59
[9] Weng Q, Lu D, Schubring J. Estimation of land surface temperature e vege-
tation abundance relationship for urban heat island studies. Remote Sens
Environ 2004;89(4):467e83.
[10] Li J, Song C, Cao L, Zhu F, Meng X, Wu J. Impacts of landscape structure on
surface urban heat islands: a case study of Shanghai, China. Remote Sens
Environ 2011;115(12):3249e63.
[11] Schwarz N, Schlink U, Franck U, Gromann K. Relationship of land surface and
air temperatures and its implications for quantifying urban heat island indi-
catorsdan application for the city of Leipzig (Germany). Ecol Indic 2012;18:
693e704.
[12] Wong NH, Chen Y. Study of green areas and urban heat island in a tropical
city. Habitat Int 2005;29(3):547e58.
[13] Yu SM, Ho KH. Planned urban industrialization and its effect on urban in-
dustrial real estate valuation: the Singapore experience. Habitat Int
2006;30(3):509e39.
[14] Han SS. Global city making in Singapore: a real estate perspective. Prog Plan
2005;64:69e175.
[15] Energy Market Authority. Energising our nation: Singapore energy statistics
2012 [Internet] [cited 2012 Dec 25]. Available from:, http://www.ema.gov.sg/
media/les/publications/EMA_SES_2012_Final.pdf; 2012 Oct.
[16] Urban Redevelopment Authority. Final report on land allocation; 2000.
[17] Peddle MT. Planned industrial and commercial developments in the United
States: a review of the history, literature, and empirical evidence regarding
industrial parks and research parks. Econ Dev Q 1993;7(1):107e24.
[18] Zhu J. Industrial globalisation and its impact on Singapore's industrial land-
scape. Habitat Int 2002;26(2):177e90.
[19] Olszewski KF, Chia LS. National development, physical planning and the
environment in Singapore. In: Chia LS, Rahman A, Dorothy Tay BH, editors.
The biophysical environment of Singapore. Singapore: Singapore University
Press; 1991. pp. 185e91.
[20] Kaushik D. Evaluation of industrial landscape in Singapore. In: 48th ISOCARP
congress 2012 [cited 2012 Dec 7]. Available from: http://www.isocarp.net/
Data/case_studies/2108.pdf; 2012.
[21] Urban Revelopment Authority. Master plan written statement [Internet]
[cited 2012 Dec 15]. Available from:, http://www.ura.gov.sg/MP2008/written_
statement.htm; 2008.
[22] Urban Redevelopment Authority. [Internet]. 2008 [cited 2013 Jan 8]. Available
from: http://www.ura.gov.sg/uramaps/?congcong_preopen.xml&preo
penMaster%20Plan.
[23] Department of Geography; National University of Singapore. Weather station
data. Available from: http://www.fas.nus.edu.sg/geog/weather/index.html.
[24] Kruger E, Givoni B. Outdoor measurements and temperature comparisons of
seven monitoring stations: preliminary studies in Curitiba, Brazil. Build En-
viron 2007;42(4):1685e98.
[25] Jusuf SK, Wong NH. Development of empirical models for an estate level air
temperature prediction in Singapore. In: Proceeding of second international
conference on countermeasures to urban heat island; 2009 Sep 21e23; Ber-
keley, California, USA. Available on: http://heatisland2009.lbl.gov/docs/
221410-jusuf-doc.pdf.
[26] Ong BL. Green plot ratio: an ecological measure for architecture and urban
planning. Landsc Urban Plan 2003;63(4):197e211.
[27] Giridharan R, Lau SSY, Ganesan S, Givoni B. Lowering the outdoor temperature
in high-rise high-density residential developments of coastal Hong Kong: the
vegetation inuence. Build Environ 2008;43(10):1583e95.
[28] Svensson MK, Eliasson I. Diurnal air temperatures in built-up areas in relation
to urban planning. Landsc Urban Plan 2002;61:37e54.
[29] Matzarakis A, Rutz F. Rayman: a tool for tourism and applied climatology. In:
Matzarakits A, Freitas CR, Scott D, editors. Developments in tourism clima-
tology. [Internet] [cited 2013 Jan 13]. Available from: http://www.mif.uni-
freiburg.de/isb/ws3/report/dTcl_2007_matazrakisrutz.pdf; 2007.
[30] Colliers International. Opportunities for industrialist in the strata-titled in-
dustrial market [White Paper]; 2012. pp. 1e9.
[31] Gallo KP, Owen TW. Satellite-based adjustments for the urban heat island
temperature bias. J Appl Meterol 1999;38:806e13.
[32] United States Environmental Protection Agency. Reducing urban heat islands:
compendium of strategies. [Internet]. [cited 2012 Aug 20]. Available from:
http://www.epa.gov/heatisld/resources/pdf/BasicsCompendium.pdf.
[33] Bradley AV, Thornes JE, Chapman L, Unwin D, Roy M. Modelling spatial and
temporal road thermal climatology in rural and urban areas using a GIS. Clim
Res 2002;22:41e55.
[34] Oke TR. Canyon geometry and the nocturnal urban heat island: comparison of
scale model and eld observations. J Clim 1981;1(3):237e54.
S.K. Jusuf et al. / Building and Environment 80 (2014) 48e60 60