Urban vegetation for reducing heat related mortality

Abstract

The potential benefit of urban vegetation in reducing heat related mortality in the city of Melbourne, Australia is investigated using a two-scale modelling approach. A meso-scale urban climate model was used to quantify the effects of ten urban vegetation schemes on the current climate in 2009 and future climates in 2030 and 2050. The indoor thermal performance of five residential buildings was then simulated using a building simulation tool with the local meso-climates associated with various urban vegetation schemes. Simulation results suggest that average seasonal summer temperatures can be reduced in the range of around 0.5 and 2 °C if the city were replaced by vegetated suburbs and parklands, respectively. With the limited buildings and local meso-climates investigated in this study, around 5-28% and 37-99% reduction in heat related mortality rate have been estimated by doubling the city's vegetation coverage and transforming the city into parklands respectively.

Full text

Urban vegetation for reducing heat related mortality
Dong Chen
a
,
*
, Xiaoming Wang
a
, Marcus Thatcher
b
, Guy Barnett
a
, Anthony Kachenko
c
,
Robert Prince
c
a
CSIRO Climate Adaptation Flagship and CSIRO Ecosystem Sciences, Melbourne, Australia
b
CSIRO Climate Adaptation Flagship and CSIRO Marine and Atmospheric Research, Melbourne, Australia
c
Nursery & Garden Industry Australia, Sydney, Australia
article info
Article history:
Received 21 January 2014
Received in revised form
1 May 2014
Accepted 3 May 2014
Available online 23 May 2014
Keywords:
Climate change
Mortality rate
Urban vegetation
Urban heat island
Heat waves
abstract
The potential benefit of urban vegetation in reducing heat related mortality in the city of Melbourne,
Australia is investigated using a two-scale modelling approach. A meso-scale urban climate model was
used to quantify the effects of ten urban vegetation schemes on the current climate in 2009 and future
climates in 2030 and 2050. The indoor thermal performance of five residential buildings was then
simulated using a building simulation tool with the local meso-climates associated with various urban
vegetation schemes. Simulation results suggest that average seasonal summer temperatures can be
reduced in the range of around 0.5 and 2

C if the city were replaced by vegetated suburbs and parklands,
respectively. With the limited buildings and local meso-climates investigated in this study, around
5e28% and 37e99% reduction in heat related mortality rate have been estimated by doubling the city’s
vegetation coverage and transforming the city into parklands respectively.
Crown Copyright Ó2014 Published by Elsevier Ltd. All rights reserved.
1. Introduction
Heat waves have been recognized as one of the major natural
hazards and kill more people than any other natural hazards in
Australia (PwC, 2011; State of Australian Cities (2013)). The heat
wave event in Melbourne, Australia during the summer of 2009
may have resulted in hundreds of excess deaths over what would
normally be expected for the period (DHS, 2009). In 2003, the
European heatwave resulted in over 70,000 excess deaths (Robine
et al., 2008). Indeed, the linkage between mortality and heat has
been long recognized (Changnon et al., 1996; Haines et al., 2006;
Luber, 2008; Huang et al., 2012). Several earlier studies have tried
to quantify the relationship between climate conditions and mor-
tality rate in Australia (Nicholls et al., 2008; Tong et al., 2010;
Loughnan et al., 2010; Tong et al., 2014). Nicholls et al. (2008)
analysed the mortality rate in Melbourne for people over 65 from
1979 to 2001. They reported that excess heat related mortality
amongst the population over 65 years of age may increase rapidly
when the mean daily temperatures (the average of yesterday’s
maximum and this morning’s minimum) exceeded about 30

C.
The impact of heat waves in major Australian cities such as
Melbourne and Sydney is likely to become worse due to further
urbanization in existing suburbs, the effects of global warming and
the ageing population. Further urbanization may potentially
intensify the urban heat island (UHI) effect, a phenomenon
whereby urbanized population centers experience warmer tem-
peratures compared with surrounding rural areas (Morris et al.,
2001; Coutts et al., 2007). At the same time, global warming pro-
jections suggest the likely increase in the number of warm nights,
heat wave frequency and duration in Australia (CSIRO, 2007;
Alexander and Arblaster, 2008). Consequently, it is essential to
develop effective strategies to protect urban communities from the
impact of heat waves.
In recent years, urban greening and cool surfaces have attracted
substantial interests and show promise in mitigating the impact of
heat waves (Rosenfeld et al.,1998; Akbari et al., 2001; Liu and Bass,
2005; Rosenzweig et al., 2006;Yu and Hien, 2006; Luber, 2008;
Alexandri and Jones, 2008; Memon et al., 2008; Bowler et al.,
2010; Wong and Lau, 2013). Urban greening can mitigate heat
wave impacts by trees that shade buildings and cool the ambient air
by evapotranspiration. Cool surfaces such as reflective roofs and
paving surfaces can further reduce heat absorption in the urban
area. Urban greening and cool surfaces can also reduce the cooling
energy demand and thus green house gas (GHG) emissions in cities
(Ca et al., 1998; Susca et al., 2011; Xu et al., 2012; Smith et al., 2012).
*Corresponding author.
E-mail address: Dong.Chen@csiro.au (D. Chen).
Contents lists available at ScienceDirect
Environmental Pollution
journal homepage: www.elsevier.com/locate/envpol
http://dx.doi.org/10.1016/j.envpol.2014.05.002
0269-7491/Crown Copyright Ó2014 Published by Elsevier Ltd. All rights reserved.
Environmental Pollution 192 (2014) 275e284

Previous researches in heat related mortality have been focused
on the relationship between the ambient weather conditions and
the mortality rate (Nicholls et al., 2008; Tong et al., 2010, 2014).
Using urban climate modelling, Kalkstein et al. (2014) investigated
the cooling effects due to different urban vegetation schemes and
surface reflectivity. Based on the relationship between the ambient
weather conditions and heat related mortality rate, they estimated
that 10% increase in the urban vegetation coverage and surface
reflectivity can result in an average 7% reduction in mortality dur-
ing heat waves in the District of Columbia. It should be noted that
heat related death occurs both indoors and outdoors. Cadot et al.
(2007) reported that 74% of excess deaths during the 2003 sum-
mer heat wave in Paris occurred among those who were living at
home. They emphasized that people most at risk from dying were
those aged over 75 years and living alone. Although there is no
available information on the locations of heat related excess deaths
in Australia, the situation in Australia may be similar to that in Paris
considering that the most vulnerable population is the elderly
people group (DHS, 2009).
Due to the importance of thermal comfort and health safety of
the living and working environment, there have been numerous
studies on thermal comfort and heat stress in buildings since early
last century (Brager and de Dear, 1998; Epstein and Moran, 2006;
ASHRAE, 2009; Djongyang et al., 2010). Based on human body
heat balance calculations and empirical observations, around 40
indices have been developed for defining thermal comfort and heat
stress (Epstein and Moran, 2006). Generally, using these heat stress
indices, thermal comfort and heat stress can be qualitatively cate-
gorized into different levels. For example, levels of heat stress may
be categorized into four groups using the Discomfort Index (DI)
which is defined as the mean of the air dry bulb and wet bulb
temperatures, i.e., DI <22, no heat stress; 22 DI <24, mild
sensation of heat; 24 DI <28, hot with difficulties in physical
work; DI 28, severe heat with risk for heat illness (Epstein and
Moran, 2006).
Qualitatively, an indoor environment with high DI index, espe-
cially for DI over 28, may potentially present high risks in causing
heat illness including heat related mortality. However, quantitative
relationships between heat related mortality rate and the DI index,
or any other heat stress index, in residential buildings are not
available. At least two reasons may attribute to the difficulties in
obtaining such relationships. First, heat stress indices are normally
developed based on the heat balance for maintaining the body-core
temperature around 37

C or empirical tests with relatively healthy
population groups such as workers in the working environment
(Epstein and Moran, 2006). However, the sensations and responses
of vulnerable population groups to heat stress are different from
healthy population groups (Hwang et al., 2007). Second, informa-
tion about the indoor thermal environment in which heat related
deaths occurred is hardly available.
Consequently, although urban greening may have the potential
in mitigating the impact of heat waves, there is currently no
established methodology in quantifying its benefit in reducing heat
related mortality in individual buildings. In this study, the role of
urban vegetation in reducing heat related mortality in the city of
Melbourne, Australia was investigated using a two-scale modelling
approach. First, a meso-scale urban climate model was used for
quantifying the effects of ten different urban vegetation schemes
on the local climate in Melbourne. Then, the indoor thermal per-
formance of five individual residential buildings was simulated
using a building simulation tool using these vegetation modified
local climates. The potential reduction in heat related mortality rate
was then estimated from the correlation between the simulated
indoor mean daily temperature and 20 year daily mortality rate
records.
2. Methodologies
2.1. Urban climate with various urban vegetation schemes
In this study, the long term average impact on the urban climate
was first established with urban climate modelling for the Mel-
bourne Central Business District (CBD) area assuming various urban
vegetation schemes. Hourly climate files for 2009, 2030 and 2050
with various urban vegetation schemes were then generated for
building thermal performance simulations using a ‘morphing’
approach.
2.1.1. Urban climate model
A recently developed urban climate model (UCM-TAPM)
(Thatcher and Hurley, 2012) was used for the investigation of the
impact of urban vegetation schemes on local urban climate. The
UCM-TAPM combines an urban canopy model (UCM) with The Air
Pollution Model (TAPM), a meso-scale climate model developed by
CSIRO (Hurley et al., 2005). The UCM includes an efficient big-leaf
model to represent in-canyon vegetation in the predominately
suburban component of Australian cities.
The model employs a multiple one-way nesting procedure to
dynamically downscale meteorological reanalyses/forecasts, typi-
cally in steps of 30 km, 10 km, 3 km and 1 km. The meteorological
component of UCM-TAPM is nested within synoptic-scale analyses/
forecasts which drive the model at the boundaries of the outer
grids. In the UCM-TAPM, a grid tile of the land surface can assign
one of 39 surface types that include a wide range of natural and
built surface types such as water body, forest, shrub land, grassland,
pasture, CBD, urban, and industrial. The characteristics of the sur-
face types such as the average building height, building height to
street canopy width ratio, vegetation coverage, leaf area index,
surface albedos etc. can be adjusted for specific urban surface
conditions. The UCM-TAPM has been validated against the mea-
surements from several urban and rural weather stations (Thatcher
and Hurley, 2012). Using UCM-TAPM, the impact of various urban
vegetation schemes on the local urban climate can be quantified.
When modelling the ‘current’Melbourne climate under various
urban forms and vegetation schemes, the downscaled reanalysis
climate data from the National Centre for Environmental Prediction
(NCEP) were used for the lateral boundaries of the outer grids for
the UCM-TAPM. For ‘future’Melbourne climates under various ur-
ban forms and vegetation schemes, UCM-TAPM simulations were
carried out with the boundary conditions based on the downscaled
climate data from GFDL2.1 under the A2 emission scenario.
2.1.2. Urban vegetation schemes
As illustrated by the green square in Fig. 1a, the Melbourne CBD
area is represented by nine 1 km 1 km grids in the UCM-TAPM.
Fig. 1b shows the corresponding map for Melbourne which covers
an area approximately 12.5 12.5 km
2
surrounding the Melbourne
CBD area. In this study, four level nesting grids were used in the
UCM-TAPM model, i.e., 30 km 30 km, 10 km 10 km,
3km3 km and 1 km 1 km grids. There are 25 25 grids for
each grid level and the total simulation domain covers an area of
750 750 km
2
with a spatial resolution of 1 km at the finest grid
level.
The impact of urban vegetation on the Melbourne CBD local
climate was investigated using the UCM-TAPM by replacing the
Melbourne CBD areas with the ten urban forms listed in Table 1. The
CBD urban form (Urban Form Number 6) represents the existing
Melbourne CBD with the vegetation and building coverage per-
centages estimated from Google images. Urban Form Number 9, i.e.,
CBD with 50% green roof, assumes 50% green roof coverage on all
the building roofs in the Melbourne CBD area. Although
D. Chen et al. / Environmental Pollution 192 (2014) 275e284276

redeveloping the Melbourne CBD area into some of the urban forms
listed in Table 1 would be unrealistic, it is important to understand
the potential effects of various urban vegetation schemes for
assisting effective and resilient future urban designs.
2.1.3. Climate file preparation
Modelling building indoor thermal performance with building
simulation tools required weather files for the local climate in
consideration. Typical Meteorological Year (TMY) weather files are
widely adopted as the reference weather files for assessing building
thermal performance. TMY weather data are specially selected so
that it represents the range of weather phenomena for a given
location. TMY weather files for the Melbourne CBD area were
derived with the weather data from or near the Melbourne
Regional Office of the Australian Bureau of Meteorology for the
period from 1976 to 2004 which is approximately centered at
around 1990 (ABCB, 2006). Future weather files for 2030 and 2050
were prepared using the ‘morphing’approach developed by
Belcher et al. (2005). With the morphing approach, weather data in
the TMY weather files were modified to take into account the
climate change effect using the flowing equations:
T¼T
0
þ
D
T
m
þ
a
Tm
ðT
0
hT
0
i
m
Þ(1)
RH ¼RH
0
þ
D
RH
m
(2)
I¼1þ
a
m;I
I
0
(3)
Fig. 1. Representation of the Melbourne CBD: (a) image from the UCM-TAPM; (b) Map of Melbourne.
Table 1
The urban forms and vegetation schemes investigated in this study.
Urban form
number
Urban form Vegetation type Vegetation
coverage (%)
Leaf area
index
Building
coverage (%)
Average building
height (m)
Building height to
canyon width Ratio
d
Irrigation
e
1 Forest (low sparse) Low sparse (woodland) 100 2.0 0 ee No
2 Shrub-land Mid dense (scrub) 100 2.6 0 ee No
3 Grassland Mid dense tussock 100 1.2 0 ee No
4 Urban (leafy) Mixed 49 3 40 6.0 0.4 Yes
5 Urban (generic) Mixed 38
a
345
a
6.0 0.4 Yes
6 CBD Mixed 15
b
365
b
12.0 1.3 Yes
7 CBD (with 1/3 Vegetation) Mixed 5 3 65 12.0 1.3 Yes
8 CBD (Double Vegetation) Mixed 33 3 62 12.0 1.3 Yes
9 CBD (50% Green Roof) Mixed 15 3
1.5
c
65 12.0 1.3 Yes
10 CBD (Double Vegetation þ50%
Green Roof)
Mixed 33 3
1.5
c
62 12.0 1.3 Yes
a
From Coutts et al. (2007).
b
Estimated from Google images. Areas other than vegetation and buildings are accounted as roads in the UCM-TAPM.
c
Leaf area index for green roof vegetation.
d
Estimated average building height divided by the street/road width.
e
Rainfall has been modelled and considered in the UCM-TAPM modelling. Irrigation is the additional watering to prevent vegetation dry out.
D. Chen et al. / Environmental Pollution 192 (2014) 275e284 277

V¼1þ
a
m;v
V
0
(4)
where T, RH, I and V are the projected future hourly dry-bulb
ambient air temperature (

C), relative humidity (%), solar irradi-
ance (W/m
2
) and wind speed (m/s) respectively;
D
represents
changes in the corresponding weather parameters projected by the
UCM-TAPM with GFDL2.1 under the A2 emission scenario;
a
m,I
and
a
m,v
are the monthly averaged percentage changes in solar irradi-
ance and wind speed (%), again projected by the UCM-TAPM with
GFDL2.1 under the A2 emission scenario; subscript 0 for the
reference climate and m for monthly-mean;
a
Tm
¼
D
TMAX
m

D
TMIN
m
hT
0max
i
m
hT
0min
i
m
;
hT
0
i
m
;hT
0max
i
m
;hT
0min
i
m
are the monthly-mean values of the
ambient dry-bulb temperature, daily maximum temperature and
daily minimum temperature from the TMY weather data, respec-
tively.
D
T
m
,
D
TMAX
m
,
D
TMIN
m
are the projected changes in the
monthly-mean values of ambient dry-bulb temperature, daily
maximum temperature and daily minimum temperature (

C)
respectively due to climate change.
When the impact of urban vegetation schemes are considered,
D
T
m
,
D
TMAX
m
and
D
TMIN
m
are obtained by the sum of the corre-
sponding temperature change due to projected climate change and
that due to different vegetation schemes as listed in Table 1. In order
to investigate the impact of various vegetation schemes during the
2009 summer in Melbourne, the weather file for the CBD urban
form from 1 July 2008 to 30 June 2009 was constructed using the
weather data obtained from the Melbourne Regional Office of the
Bureau of Meteorology. This weather file was then modified using
the
D
T
m
,
D
TMAX
m
and
D
TMIN
m
obtained for different urban vege-
tation schemes by the UCM-TAPM simulations which will be
further discussed in Section 3.1.
2.2. Indoor thermal environment and mortality
2.2.1. Mortality and population data
Mortality data from 1988 to 2007 for Melbourne were obtained
from the Health and Vitals Statistics Unit, Australian Bureau of
Statistics (ABS). Due to privacy protection, the data obtained from
ABS were limited for the Melbourne Statistical Division (SD) of
usual residence by sex and by two age groups, being 0e75 and 75þ.
Melbourne SD covers the metropolitan area of Melbourne as well as
its surrounding urban fringe and rural areas, which including the
Dandenong Ranges, the Yarra Valley and the Mornington Peninsula.
It has a population exceeding 3.5 million and accounts for over 70%
of the population of the state of Victoria. In order to calculate the
mortality rate, the corresponding populations by sex bythe two age
groups from 1988 to 2007 in Melbourne SD were also obtained
from ABS. In this study, constrained by the availability of mortality
data, the average mortality rate obtained for Melbourne SD is taken
to be the mortality rate for the city of Melbourne. It is understood
that this assumption may reduce the accuracy of the current study.
It is believed that using the same methodology adopted in this
study, the accuracy of the findings presented in this research can be
further improved if granular mortality data for the city of Mel-
bourne is available in the future.
2.2.2. Building thermal performance
In this study, the thermal performance of five residential
buildings was modelled by AccuRate ea residential house energy
rating software used in Australia (Delsante, 2005). AccuRate was
developed by coupling a frequency response building thermal
model (Walsh and Delsante, 1983) and a multi-zone ventilation
model (Ren and Chen, 2010) for thermal performance and energy
requirement calculation of residential buildings. The five residen-
tial buildings are assumed to operate without heating and air
conditioning. It is also assumed that the occupants operate the
windows and doors based on the following rules:
Windows and doors are closed if indoor air temperature is
below 22

C;
If indoor air temperature is above 24

C and outdoor air tem-
perature is below indoor air temperature, windows and doors
are opened. Otherwise, windows and doors are closed.
In Melbourne, detached houses represent around 76% of the
residential housing stock and the remaining 24% consists of
buildings such as semi-detached buildings, flats, units, apartments
(ABS, 2011). In this study, five residential buildings were used
which include three detached houses, a semi-detached three
bedroom two-storey townhouse and a two bedroom apartment at
the top of a two-storey building. It is assumed that there was no
insulation in these buildings to represent the low-end housing
stock in Melbourne with their occupants potentially exposed to
high heat stress risks during heat weave periods. Table 2 describes
the design specifications of the five buildings.
2.2.3. Correlate indoor mean air temperature with mortality rate
In order to understand the linkage between indoor air temper-
ature and mortality rate in Melbourne, hourly simulations were
carried out for 20 years from 1 January 1988 to 31 December 2007
for the five buildings facing north, east, south and west. The hourly
weather data used for the 20 year simulations were based on the
records from the Bureau of Meteorology.
Considering that occupants are normally in the living room
during daytime and in the bedroom at nighttime, the mean daily
temperature, T
m, daily
, for a building is defined as the average of
yesterday’s daytime (after 7am) maximum in the living room and
this morning’s (before 7am) minimum in the master bedroom. In
the 20 years from 1 January 1988 to 31 December 2007, there are a
total of 7305 mean daily temperatures for each building grouped
into consecutive 0.5

C bands. The average mortality rates for the
city of Melbourne corresponding to a particular mean daily tem-
perature band were obtained from the mortality rate of the days
when the mean daily temperatures sit in a particular 0.5

C mean
daily temperature band. For example, the average mortality rate
Table 2
Brief descriptions of the five buildings investigated in this study.
Number of
bedroom
Floor area (m
2
) External wall Internal walls Floor Ceiling Windows
House1 4 161 Brick veneer Plasterboard on studs Concrete slab Plasterboard Timber, single-glazed
House2 3 88 Brick Veneer Plasterboard on studs Concrete slab Plasterboard Aluminium, single-glazed
House3 3 88 Brick Veneer Plasterboard on studs Timber floor to subfloor Plasterboard Aluminium, single-glazed
Townhouse 3 108 Brick veneer and
weather board
Plasterboard on studs Concrete slab Plasterboard Aluminium, single-glazed
Apartment 2 64 Brick veneer Plasterboard on studs Timber floor to ground
floor underneath
Plasterboard Aluminium, single-glazed
D. Chen et al. / Environmental Pollution 192 (2014) 275e284278

corresponding to the mean daily temperature band from 28

Cto
28.5

C is the average of the mortality rates for the city of Mel-
bourne for all the days (in the 20 years) when the building has the
mean daily temperature between 28

C and 28.5

C.
The relationships between the mean daily temperatures and
the average mortality rate can be established from the 20 year
daily mortality rate and the simulated daily mean temperature
for the five buildings in four directions for both males and fe-
males. This approach for correlating mortality rate with the in-
door mean daily temperature is similar to those used by several
researchers (Donaldson et al., 2003;McMichael et al., 2006;To ng
et al., 2014) except that the correlations established in previous
studies were between mortality rate and the ambient air mean
daily temperature. These correlations between the indoor mean
daily temperatures and the average mortality rate were used
to estimate the mortality rate in 2009, 2030 and 2050 in
Section 3.2.
3. Results and discussions
3.1. The impact of urban vegetation on local climate
UCM-TAPM simulations were carried out for the Melbourne CBD
area in 2001, 2009, 2030, 2047, 2050, 2087 and 2090 with the ten
urban forms as described in Table 1. It is noted that like other urban
climate models, the UCM-TAPM is used for predicting urban climate
probability distributions and seasonal variability of the climatic
variables, especially when UCM-TAPM is nested in Global Climate
Models predicting global warming scenarios rather than reanalyses
of observed weather conditions (Thatcher and Hurley, 2012). The
UCM-TAPM is not intended for predicting short term climate phe-
nomena such as the hourly air temperature or hourly wind speed.
In this study, the impact of a specific urban form on the local
climate in the Melbourne CBD area is expressed as the changes in
three summer seasonal air temperatures obtained by the UCM-
TAPM simulations for this particular urban form relative to that
for the CBD urban form (Urban Form Number 6), i.e.,
D
T
Mean
,
D
T
Min
and
D
T
Max.
Here,
T
Min
: the average summer daily minimum temperature (average
over January, February and December);
T
Mean
: the average summer mean temperature; and
T
Max
: the average summer daily maximum temperature.
Fig. 2aec shows the
D
T
Min
,
D
T
Mean
and
D
T
Max
obtained by the
UCM-TAPM simulations for various urban forms in ‘current’years in
2001 and 2009 using the NECP reanalysis climate data and for five
future years in 2030, 2047, 2050, 2087 and 2090 projected using
GFDL2.1. In Fig. 2, DV and GR are the abbreviations for double
vegetation and green roof respectively. The thick green lines are the
mean values of the
D
T
Min
,
D
T
Mean
and
D
T
Max
for each urban form
across the seven years. It is seen that the average summer daily
maximum temperature in the Melbourne CBD area reduces with
the increase in urban vegetation coverage with the generic urban
area predicted to be around 0.5

C cooler than the existing CBD
urban forms. The maximum cooling potential of around 2

C could
be achieved if the CBD area would have been transformed to a
natural forest park land. Similar cooling potential for urban vege-
tation was observed for the corresponding T
Min
and T
Mean
in the
CBD area.
It was found that despite the projected warming trends in the
future, for a given urban vegetation scheme, the variations in
D
T
Min
,
D
T
Mean
and
D
T
Max
are generally within 20% of the means
for the seven years. The cooling effect of urban vegetation is
determined by two aspects, vegetation shading and
Fig. 2. Projections of the
D
T
Min
,
D
T
Mean
and
D
T
Max
for the ten urban forms in 2001,
2009, 2030, 2047, 2050, 2087 and 2090: (a)
D
T
Min
(C); (b)
D
T
Mean
(C); (c)
D
T
Max
(C).
D. Chen et al. / Environmental Pollution 192 (2014) 275e284 279

evapotranspiration, with the latter the dominant effect since the
vegetation shading effect for a specific urban form is more or less
fixed. Evapotranspiration rate of urban vegetation scenarios can
vary due to variation in the rainfall in each year. However, as
shown in Fig. 2, overall, these variations in vegetation shading and
evapotranspiration have limited effect on the
D
T
Min
,
D
T
Mean
and
D
T
Max
for a specific vegetation scheme. Consequently, as
mentioned in Section 2, the weather files for a particular year with
a specific urban vegetation scheme can be constructed using the
‘morphing’approach with the
D
T
m
,
D
TMAX
m
and
D
TMIN
m
ob-
tained by the UCM-TAPM. In this study, the
D
T
m
,
D
TMAX
m
and
D
TMIN
m
for a specific urban vegetation scheme were the average
values obtained from the seven year UCM-TAPM simulation re-
sults. With this approach, the constructed weather file retains the
long term average temperature changes due to a specific urban
vegetation scheme and projected climate change.
3.2. Excess mortality rate due to indoor heat stress
As discussed in Section 2.2.3, 20 relationships between the
mean daily temperatures and the average mortality rate can be
established from 20 year daily mortality rate and the simulated
daily mean temperature for the five building in four directions for
both males and females. Fig. 3 shows the relationships between the
mean daily temperature, T
m,daily
, and the average mortality rates for
males and females over 75 years old for House1 facing north and
south respectively. Similar to the “U”shape relationship observed
between the mortality rate and the mean daily ambient air tem-
peratures by many previous studies (Donaldson et al., 2003;
McMichael et al., 2006;Tong et al., 2014), extremely cold and hot
indoor mean daily temperatures in buildings also correspond to
high average mortality rates. Similar trends were observed for the
remaining buildings with different facing directions. For each set of
the relationship between the average mortality rates and the mean
daily indoor temperatures, a correlation was obtained using the R
Software (version 3.0.2) as shown in Fig. 3. Consequently, a total of
20 correlations were obtained for males and females over 75 years
old respectively for these five buildings in four directions. These
correlations are used to estimate the mortality rate in 2009, 2030
and 2050 in the following.
It is noted that the correlations obtained as shown in Fig. 3 are
the long term average relationships between the mean daily tem-
perature, T
m,daily
, and the average mortality rates and may not be
applicable to a specific heat wave event. It is also noted that each
plot in Fig. 3 was obtained by assuming that the Melbourne CBD
area has only one type of building which faces a single direction.
This assumption does not adequately reflect the present urban
environment in Melbourne. Consequently, Fig. 3 should be
interpreted with care as it does not necessary suggest that a
particular mean daily temperature in a building with a particular
facing direction will cause the corresponding mortality rate in this
building. Nevertheless, Fig. 3 suggests that the high mortality rate
during a heat wave period may correlate with the high mean daily
temperatures in various individual buildings.
From Fig. 3 for House1 and results for other four buildings, it was
found that the average mortality rate increases rapidly when the
mean daily indoor temperature T
m,daily
is approximately 28.5

C.
Therefore, in this study, the mortality rate at T
m,daily
¼28.5

Cis
assumed as the base mortality rate. The excess mortality rate due to
heat stress was then estimated as the sum of the mortality rate for
those days with mean daily temperature of the building above
28.5

C minus the base mortality rate. For example, the excess
mortality rate for the north facing House1 for a specific urban
vegetation form is calculated as:
Here, MR
excess;House1;north;veg scheme
is the excess heat related
mortality rate for a north facing House1 for a specific urban vege-
tation scheme. Days
T
m;daily
is the number of days when the mean
daily temperature T
m,daily
is in a 0.5

C temperature band above
28.5

C. The estimated excess mortality rate for each building type
is obtained as the average of the four facing directions of this
building type, e.g., the excess mortality rate for House1 is calculated
as:
MR
excess;House1;veg scheme
¼1
4MR
excess;House1;north;veg scheme
þMR
excess;House1;east;veg scheme
þMR
excess;House1;south;veg scheme
þMR
excess;House1;west;veg scheme

(6)
The potential impact on excess heat related mortality rate for a
specific urban vegetation scheme was then calculated as the per-
centage difference of the excess mortality rate in comparison with
that for the CBD urban scheme as follows:
This potential impact on excess mortality rate may be consid-
ered as an estimation of the difference in the heat related mortality
rate when the Melbourne CBD area has a specific urban vegetation
scheme relative to the CBD scheme.
Fig. 4 show the potential impact on excess mortality rate for
males and females over 75 respectively with different urban
vegetation schemes in 2009, 2030 and 2050 for each of the five
buildings. Although there are variations among different buildings
across different years, the overall trends are consistent in that ur-
ban vegetation can potentially reduce the excess mortality rate due
MR
excess;House1;north;veg scheme
¼X
T
m;daily
>28:5

C
hMR
T
m;daily
;House1;north;veg scheme
MR
T
m;daily
¼28:5
o
C;House1;north;veg scheme
Days
T
m;daily
i(5)
MR
impact;House1veg scheme
¼MR
excess;House1;veg scheme
MR
excess;House1;CBD

MR
excess;House1;CBD
100% (7)
D. Chen et al. / Environmental Pollution 192 (2014) 275e284280

to heat stress. In general, the reduction in excess mortality rate
increases with an increase in vegetation coverage and intensity
with the forest scheme (assuming the whole Melbourne CBD area
has a forest scheme) achieving the best performance from 37% up to
99% reduction in the excess mortality rate.
Results for the five buildings show that increasing the Mel-
bourne CBD vegetation coverage from 15% to 33% may reduce the
average heat related mortality rate in the range between 5% and
28%. It is understood that there is no similar research in Melbourne
for the benefit of green coverage on heat related mortality. How-
ever, this predicted reduction in the average heat related mortality
rate is comparable with the results presented in Kalkstein et al.
(2014), who estimated that 10% increase in the urban vegetation
coverage and surface reflectivity can result in an average 7%
reduction in heat related mortality during heat wavesin the District
of Columbia.
Fig. 3. Relationships between the mean daily temperature in House1 and the average mortality rate in Melbourne from 1 January 1988 to 31 December 2007: (a) North facing,
males; (b) North facing, females; (c) South facing, males; (d) South facing, females.
D. Chen et al. / Environmental Pollution 192 (2014) 275e284 281

-110%
-90%
-70%
-50%
-30%
-10%
10%
30%
ReducƟon in mortality Rate (%)
2009 House1
2009 House2
2009 House3
2009 TownHouse
2009 Apartment
-110%
-90%
-70%
-50%
-30%
-10%
10%
30%
ReducƟon in mortality Rate (%)
2009 House1
2009 House2
2009 House3
2009 TownHouse
2009 Apartment
(a) (b)
-110%
-90%
-70%
-50%
-30%
-10%
10%
30%
ReducƟon in mortality Rate (%)
2030 House1
2030 House2
2030 House3
2030 TownHouse
2030 Apartment
-110%
-90%
-70%
-50%
-30%
-10%
10%
30%
ReducƟon in mortality Rate (%)
2030 House1
2030 House2
2030 House3
2030 TownHouse
2030 Apartment
(c) (d)
-110%
-90%
-70%
-50%
-30%
-10%
10%
30%
ReducƟon in mortality Rate (%)
2050 House1
2050 House2
2050 House3
2050 TownHouse
2050 Apartment
-110%
-90%
-70%
-50%
-30%
-10%
10%
30%
ReducƟon in mortality Rate (%)
2050 House1
2050 House2
2050 House3
2050 TownHouse
2050 Apartment
(e) (f)
Fig. 4. Estimation of the reduction in heat related mortality rate with various urban vegetation schemes for: (a) males in 2009; (b) females in 2009; (c) males in 2030; (d) females in
2030; (e) males in 2050; (f) females in 2050.
D. Chen et al. / Environmental Pollution 192 (2014) 275e284282

It was also found that the reduction in the mortality rate for
males is only slightly different from that for females for a given
urban vegetation scheme in a specific building. This is due to two
reasons. First, in a given building, both the male and female occu-
pants are assumed to expose to the same mean daily temperature.
Second, from Eqs. (1)e(3), it can be shown that for a specific house
in a given urban vegetation scheme, if the relationship between
mortality rate and the mean daily temperature is exactly linear
with a gradient of k, the potential impact on excess heat related
mortality rate obtained by Eq. (3) is only dependent on the number
of days with T
m,daily
>28.5

C for this urban vegetation scheme
and for the CBD scheme. The potential impact on excess heat
related mortality rate obtained by Eq. (3) does not change with the
gradient k.
As shown in Fig. 3, the relationship between mortality rate and
the mean daily temperature is essentially linear for both males and
females after T
m,daily
¼28.5

C. Consequently, the percentage re-
ductions in the mortality rate for a given building are approxi-
mately the same for males and females for a given urban vegetation
scheme in a specific year in comparison with the CBD vegetation
scheme. However, the percentage reductions in the mortality rate
for both male and female vary among different buildings or the
same building in different years, since different buildings and
different years (thus different weather conditions) result in
different number of days with T
m,daily
>28.5

C for a given urban
vegetation scheme and for the CBD vegetation scheme. This finding
suggests that urban greening may have similar benefit in reducing
heat related mortality rate for both males and females.
Considering that it is a first attempt in quantifying urban
vegetation in mitigating heat related mortality rate at the scale of
buildings, the current study may be improved in several aspects in
the future. First, there are uncertainties involved in future climate
projections and thus the future weather files used for building
thermal performance in this study. This may be improved in the
future by using multiple Atmosphere-Ocean General Circulation
Models (AOGCMs) with different carbon emissions scenarios to
take into account the uncertainties of individual AOGCM models as
recommended by the Intergovernmental Panel on Climate Change
(IPCC, 2007). Second, in this study, the average mortality rate ob-
tained for Melbourne SD is taken to be the mortality rate for the city
of Melbourne. Using the same methodology, the results can be
improved if granular mortality data for the city of Melbourne can be
available in the future. Third, uncertainty analysis in the results may
be carried out by considering the confidence level in the correla-
tions between the mean daily temperature, T
m,daily
, and the average
mortality rates. Fourth, as discussed in Section 3.1, the building
sample size in this study is small and the findings from this study
will need to be further verified with large building sample sizes.
Nevertheless, the research findings are encouraging and show the
potential benefit of urban vegetation in reducing the heat related
excess mortality rate by mitigating the impact of heat waves and
projected global warming.
4. Conclusions
The potential for urban vegetation in reducing heat related
mortality for 2009 and the projected future climates in 2030 and
2050 in the city of Melbourne, Australia was investigated for ten
different urban vegetation schemes using a two-scale modelling
approach: a meso-scale urban climate model and a building ther-
mal performance model. Simulation results showed that the
average seasonal summer temperatures can be reduced in the
range of around 0.5 and 2

C if Melbourne CBD were replaced by
vegetated suburbs and planted parklands, respectively. It was also
found that despite the projected warming in the future, the average
seasonal cooling potential due to various urban vegetation schemes
remains reasonably unchanged in comparison with those predicted
for the current climate, indicating little dependency on climate
change.
Simulations of the indoor thermal environment for five resi-
dential buildings in 2009 and for projected future climates in 2030
and 2050 showed that urban vegetation can potentially reduce
excess mortality rate in the city of Melbourne due to heat stress. An
increase in the Melbourne CBD vegetation coverage from 15% to
33% may reduce the average heat related mortality rate in the range
between 5% and 28%. Reduction in the excess mortality rate in the
range from 37% up to 99% is estimated by replacing the whole
Melbourne CBD area with forest parkland. Results also suggest that
urban greening may have similar benefit in the percentage reduc-
tion in heat related mortality rate for both males and females.
Although variations were predicted among different buildings and
in different years, these findings may provide confidence as well as
a methodology for urban planners in mitigating urban heat island
and heat wave impact with urban greening strategies.
The modelling approach used in this study is a first attempt in
quantifying urban vegetation in mitigating heat related mortality
rate at the scale of buildings. As discussed in Section 3.2, the current
methodology can be further improved in at least the following four
aspects in the future: take into account the uncertainties involved
in future climate projections by using multiple AOGCMs with
different carbon emissions scenarios; take into account the confi-
dence level of correlation between the mean daily temperature and
the average mortality rates; use granular mortality data for the city
of Melbourne for establishing the correlation between the mor-
tality rate and indoor thermal environment; and, extend the
building sample size. Further studies using the two-scale modelling
approach is also required to validate our results.
Acknowledgement
This study was partly funded by the Horticulture Australia
Limited using the Nursery Industry Levy (Project # NY11013 and
NY12018) and CSIRO Climate Adaptation Flagship. The authors
would like to thank Mr Zhiwei Xu in Queensland University of
Technology on using R software and Mr Felix Lipkin in CSIRO for
providing the map of Melbourne.
References
ABCB, 2006. Protocol for House Energy Rating Software. Australian Building Codes
Board.
Akbari, H., Pomerantz, M., Taha, H., 2001. Cool surfaces and shade trees to reduce
energy use and improve air quality in urban areas. Sol. Energy 70, 295e310.
Alexander, L.V., Arblaster, J., 2008. Assessing trends in observed and modelled
climate extremes over Australia in relation to future projections. Int. J. Climatol.
29 (3), 417e435.
Alexandri, E., Jones, P., 2008. Temperature decreases in an urban canyon due to
green walls and green roofs in diverse climates. Build. Environ. 43, 480e493.
ASHRAE Handbook, 2009. 2009 Fundamentals. American Society of Heating,
Refrigerating and Air-Conditioning Engineers, Inc.
Australia, 2011. Census, Australian Bureau of Statistics (ABS). http://www.abs.gov.au
(accessed in January 2014).
Belcher, S.E., Hacker, J.N., Powell, D.S., 2005. Constructing design weather data for
future climates. Build. Serv. Eng. Res. Technol. 26 (1), 49e61.
Bowler, D.E., Buyung-Ali, L., Knight, T.M., Pullin, A.S., 2010. Urban greening to cool
towns and cities: a systematic review of the empirical evidence. Landsc. Urban
Plan. 97, 147e155.
Brager, G.S., de Dear, R., 1998. Thermal adaptation in the built environment: a
literature review. Energy Build. 27, 83e96.
Ca, V.T., Asaeda, T., Abu, E.M., 1998. Reductions in air conditioning energy caused by
a nearby park. Energy Build. 29, 83e92.
Cadot, E., Rodwin, V.G., Spira, A., 2007. In the heat of the summer: lessons from the
heat waves in Paris. J. Urban Health 84 (4), 466e468.
Changnon, S.A., Kunkel, K.E., Reinke, B.C., 1996. Impacts and responses to the 1995
heat wave: a call to action. Bull. Am. Meteorol. Soc. 77, 1497e1505.
D. Chen et al. / Environmental Pollution 192 (2014) 275e284 283

Coutts, A., Beringer, J., Tapper, N., 2007. Impact of increasing urban density on local
climate: spatial and temporal variations in the surface energy balance in Mel-
bourne, Australia. J. Appl. Meteorol. Climatol. 46, 477e493.
CSIRO, 2007. Climate Change in Australia eTechnical Report.
Delsante, A.E., 2005. Is the new generation of building energy rating software up to
the task? ea review of AccuRate. In: ABCB Conference ‘Building Australia’s
Future 2005’, Surfers Paradise, Australia, 11e15 September 2005.
DHS, January 2009. Heatwave in Victoria: an Assessment of Health Impacts.
Victorian Government Department of Human Services, Victoria, 2009. http://
www.health.vic.gov.au/chiefhealthofficer/publications/heatwave.htm (accessed
in January 2014).
Djongyang, N.I., Tchinda, R., Njomo, D., 2010. Thermal comfort: a review paper.
Renew. Sustain. Energy Rev. 14, 2626e2640.
Donaldson, G.C., Keatinge, W.R., Nayha, S., 2003. Changes in summer temperature
and heat related mortality since 1971 in North Carolina, South Finland, and
Southeast England. Environ. Res. 91, 1e7.
Epstein, Y., Moran, D., 2006. Thermal comfort and the heat stress indices. Ind.
Health 44, 389e398.
Haines, A., Kovats, R.S., Campbell-Lendrum, D., Corvalan, C., 2006. Climate change
and human health: impacts, vulnerability, and mitigation. Lancet 367 (9528),
2101e2109.
Huang, C., Barnett, A.G., Wang, X., Tong, S., 2012. Effects of extreme temperatures on
years of life lost for cardiovascular deaths: a time series study in Brisbane,
Australia. Circul.: Cardiovasc. Qual. Outcomes 5 (5), 609e614.
Hurley, P., Physick, W., Luhar, A., 2005. TAPM ea practical approach to prognostic
meteorological and air pollution modeling. Environ. Model Softw. 20, 737e752.
Hwang, R.L., Lin, T.P., Cheng, M.J., Chien, J.H., 2007. Patient thermal comfort
requirement for hospital environments in Taiwan. Build. Environ. 42, 2980e
2987.
IPCC (Intergovernmental Panel on Climate Change), 2007. Climate Change 2007 e
the Fourth Assessment Report. Cambridge University.
Kalkstein, L.S., Sailor, D., Shickman, K., Sheridan, S., Vanos, J., 2014. Assessing the
Health Impacts of Urban Heat Island Reduction Strategies in the District of
Columbia. http://www.coolrooftoolkit.org/wp-content/uploads/2013/10/DC-
Heat-Mortality-Study-for-DDOE-FINAL.pdf (accessed in January 2014).
Liu, K., Bass, B., 2005. Performance of Green Roof Systems. Report No. NRCC-47705.
National Research Council Canada, Toronto, Canada.
Loughnan, M., Nicholls, N., Tapper, N., 2010. Mortalityetemperature thresholds for
ten major population centres in rural Victoria, Australia. Health & Place 16,
1287e1290.
Luber, G., 2008. Climate change and extreme heat events. Am. J. Prev. Med. 35 (5),
429e435.
McMichael, A.J., Woodruff, R.E., Hales, S., 2006. Climate change and human health:
present and future risks. Lancet 367, 859e869.
Memon, R.A., Leung, D.Y.C., Liu, C.H., 2008. A review on the generation, determi-
nation and mitigation of Urban Heat Island. J. Environ. Sci. 20, 120e128.
Morris, C.J.G., Simmonds, I., Plummer, N., 2001. Quantification of the influences of
wind and cloud on the nocturnal urban heat island of a large city. J. Appl.
Meteorol. 40, 169e182.
Nicholls, N., Skinner, C., Loughnan, M., Tapper, N., 2008. A simple heat alert system
for Melbourne, Australia. Int. J. Biometeorol. 52, 375e384.
PwC report, Nov. 2011. Protecting Human Health and Safety During Severe and
Extreme Heat Events. PricewaterhouseCoopers Australia (PwC) Report.
Ren, Z., Chen, D., 2010. Enhanced air flow modelling for AccuRate ea nationwide
house energy rating tool in Australia. Build. Environ. 45, 1276e1286.
Robine, J.M., Cheung, S.L.K., Roy, S.L., Oyen, H.V., Griffiths, C., Michel, J.P.,
Herrmann, F.R., 2008. Death toll exceeded 70,000 in Europe during the summer
of 2003. C. R. Biol. 331, 171e178.
Rosenfeld, A.H., Akbari, H., Romm, J.J., Pomerantz, M., 1998. Cool communities: stra-
tegies for heat island mitigation and smog reduction. Energy Build. 28, 51e62.
Rosenzweig, C., Solecki, W., Parshall, L., Gaffin, S., Lynn, B., Goldberg, R., Cox, J.,
Hodges, S., 2006. Mitigating New York City’s heat island with urban Forestry,
living roofs, and Light surfaces. In: Sixth Symposium on the Urban Environment
and Forum on Managing our Physical and Natural Resources, American Mete-
orological Society, January 31, 2006, Atlanta, GA.
Smith, G.B., Aguilar, J.L.C., Gentle, A.R., Chen, D., 2012. Multi-parameter sensitivity
analysis: a design methodology applied to energy efficiency in temperate
climate houses. Energy Build. 55, 668e673.
State of Australian Cities, 2013. Major Urban Unit. Department of Infrastructure and
Transport, 2013.
Susca, T., Gaffin, S.R., Dell’Osso, G.R., 2011. Positive effects of vegetation: urban heat
island and green roofs. Environ. Pollut. 159, 2119e2126 .
Thatcher, M., Hurley, P., 2012. Simulating Australian urban climate in a mesoscale
atmospheric numerical model. Boundary-Layer Meteorol. 142, 149e175.
Tong, S., Ren, C., Becker, N., 2010. Excess deaths during the 2004 heatwave in
Brisbane, Australia. J. Biometeorol. 54 (4), 393e400.
Tong, S.L., Wang, X.Y., Yu, W.W., Chen, D., Wang, X.M., 2014. The impact of heat-
waves on mortality in Australia: a multi-city study. Br. Med. J. Open. 4, e003579
http://dx.doi.org/10.1136/bmjopen-2013-003579.
Walsh, P.J., Delsante, A., 1983. Calculation of the thermal behaviour of multi-zone
buildings. Energy Build. 5, 231e242.
Wong, J.K.W., Lau, L.S.K., 2013. From the ‘urban heat island’to the ‘green island’?A
preliminary investigation into the potential of retrofitting green roofs in
Mongkok district of Hong Kong. Habitat Int. 39, 25e35.
Xu, T., Sathaye, J., Akbari, H., Garg, V., Tetali, S., 2012. Quantifying the direct benefits
of cool roofs in an urban setting: reduced cooling energy use and lowered
greenhouse gas emissions. Build. Environ. 48, 1e6.
Yu, C., Hien, W.N., 2006. Thermal benefits of city Parks. Energy Build. 38 (2), 105e120.
D. Chen et al. / Environmental Pollution 192 (2014) 275e284284