ARTICLE IN PRESS

Building and Environment 43 (2008) 480493

www.elsevier.com/locate/buildenv

Temperature decreases in an urban canyon due to green walls and green

roofs in diverse climates
Eleftheria Alexandria, Phil Jonesb,
a
Mantzakou 2-6, 114 73 Athens, Greece
b
Welsh School of Architecture, Cardiff University, King Edward VII Avenue, Cardiff CF10 3NB, UK
Received 23 February 2006; received in revised form 24 July 2006; accepted 31 October 2006

Abstract

This paper discusses the thermal effect of covering the building envelope with vegetation on the microclimate in the built environment,
for various climates and urban canyon geometries. A two-dimensional, prognostic, micro scale model has been used, developed for the
purposes of this study. The climatic characteristics of nine cities, three urban canyon geometries, two canyon orientations and two wind
directions are examined. The thermal effect of green roofs and green walls on the built environment is examined in both inside the canyon
and at roof level. The effects of this temperature decrease on outdoors thermal comfort and energy savings are examined. Conclusions
are drawn on whether plants on the building envelope can be used to tackle the heat island effect, depending on all these parameters
taken into consideration.
r 2006 Elsevier Ltd. All rights reserved.

Keywords: Green roofs; Green walls; Urban canyon; Plants; Built environment

1. Introduction These changes have a direct effect on the local climate of

Since the beginning of human existence man has clearly
intended to alter his microclimate, to a more human-
friendly one, protecting himself from extreme climatic
conditions. Even from the rst evidence of Neolithic houses
and settlements, it is obvious that they were not sited in a
purely natural environment, but in a part of nature
transformed according to a human plan [1]. With the
evolution of human societies, settlements were trans-
formed, evolved into villages, towns or cities, developed
or faded away, according to the geographical, economic,
social and cultural transformations taking place through-
out time. With the Industrial Revolution, urban spaces
expanded dramatically, much faster and with much more
signicant changes than in their previous evolutionary
periods. The large areas modern cities occupy, their
structure, materials and the general lack of vegetation
cannot but have altered the climatic characteristics of
urban spaces.
Corresponding author.
urban spaces, especially the central parts of the city,
causing a signicant rise of the urban temperature and
other alterations, known as the heat island effect. This may
cause serious local climatic unpleasant conditions and even
imperil human health, especially for cities in climates with a
distinctively hot season [2,3]. The moderation of extreme
heat in the local environment of such climates could mean
not only their sustainability, but also the potential of
occupying them without the morbidity and mortality risks
caused by excessive heat [4,5].
On prima facie evidence, the general lack of vegetation in
existing cities is one of the factors affecting the formation
of raised urban temperatures. In most urban spaces,
appreciable amounts of vegetation exist mostly concen-
trated in parks or recreational spaces. Although parks
manage to lower temperatures within their vicinity [69],
they are incapable of thermally affecting the concentrated
built spaces where people live, work and spend most of
their urban lives. By placing vegetation within the built
space of the urban fabric, raised urban temperatures can
decrease within the human habitats themselves and not
only in the detached spaces of parks. Urban surfaces which

0360-1323/$ - see front matter r 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.buildenv.2006.10.055
 ARTICLE IN PRESS
E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493 481

Nomenclature qE heat gains/losses from the buildings fabric

(W/m2)
Latin Letters t time (s)
T air temperature (K)
C net sensible heat loss on leaf tissue (W/m2) Tin indoors air temperature (K)
c 1: isobaric specic heat of component 1 (moisture) T leaf surface temperature (K)
of the mixture (J/kg K) Tout outdoors air temperature (K)
c 2: isobaric specic heat of component 2 (air) of U average buildings fabric U-value (W/m2 K)
the mixture (J/kg K) u air velocity in x axis (m/s)
cc building material specic heat capacity (J/kg K) w air velocity in z axis (m/s)
cpa isobaric specic heat capacity of air (J/kg K)
cp specic heat capacity of the leaf tissue (J/kg K) Greek Letters
D binary diffusion coefcient (m2/s)
I1 any source of mass of moisture (kg/m3s) aa coefcient of thermal diffusivity of air (m2/s)
KHx eddy diffusion coefcient of energy in x-axis ac building material thermal diffusion coefcient
(m2/s) (m2/s)
KHz eddy diffusion coefcient of energy in z-axis am diffusion coefcient of moisture in the building
(m2/s) material (m2/s)
KEx eddy diffusion coefcient of water vapour in x- e evaporation number of the building material
axis (m2/s) l latent heat of vaporisation (J/kg)
KEz eddy diffusion coefcient of water vapour in z- lE net latent heat loss on leaf tissue (W/m2)
axis (m2/s) ra density of moist air (kg/m3)
q relative concentration of water vapour, ex- r density of the leaf tissue (kg/m3)
pressed as specic humidity (kg/kg) Fn net heat gain on leaf tissue from radiation (W/m2)

are not used, such as the building envelope (walls and urban
roofs), could easily be covered with vegetation and alter the
microclimate of the built environment, as well as the local
10m

climate of the city. The magnitude of temperature

urban urban
decreases due to this transformation depends on the
climatic characteristics, the amount of vegetation and
urban geometry.
This paper presents the results of a quantitative research
H

on how the heat island effect can be tackled by covering the in in

envelope of urban buildings with vegetation. The aim of ground a1
this research has been to assess the potential of mitigating W a2 in
raised urban temperatures through vegetation, for different
urban geometries and climates. Fig. 1. Two-dimensional canyon model.

heat transfer [11,12]:

2. Methodology  
dT l c1  c2
raa rT I1  DrqrT, (1)
dt cpa ra cpa
A two-dimensional, prognostic (dynamic) micro-scale
model has been developed and programmed in C++,
dq 1
describing heat and mass transfer in a typical urban rDrq I 1 , (2)
dt ra
canyon (Fig. 1). The differential equations describing
heat and mass transfer in the air, building materials where T is the air temperature (in K), t is time (in s), aa the
(considered as capillaryporous bodies), soil and coefcient of thermal diffusivity of air (in m2/s), D the
vegetation have been solved with nite differences binary diffusion coefcient (in m2/s), cpa the isobaric
approximations, where surface elements are replaced specic heat capacity of air (in J/kg K), ra is the density
by nodes [10]. The effect of vapour gradients on of moist air (in kg/m3), l the latent heat of vaporisation (in
temperature gradients has been described analytically in J/kg), I1 is any source of mass of moisture (in kg/m3s), q is
the air nodes near the surfaces as the effect of diffusion on the relative concentration of water vapour, expressed as
 ARTICLE IN PRESS
482 E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493

specic humidity (in kg/kg), c1 is the isobaric specic heat Regarding plants, they are considered to be a layer
of component 1 (moisture) of the mixture (in J/kg K) and c2 consisting of canopy leaves and the air among them.
the isobaric specic heat of component 2 (air) of the Equations describing heat and mass transfer in the air are
mixture (in J/kg K). According to Eckert and Drake [12], the ones given by Eqs. (1) and (2), while heat transfer in the
the effect of thermal diffusion on heat transfer (fourth term leaf is given by
of Eq. (1)) can be neglected in normal engineering mass- dT
transfer processes. However, they point out [12] that it r cp Fn  C  lE, (7)
dt
contributes essentially when temperature gradients are
extremely large. In the description of the thermal ex- where r is the density of the leaf tissue (in kg/m3), cp is
changes in the built environment, where temperature the specic heat capacity of the leaf tissue (in J/kg K), T
gradients in the boundary layer of surfaces exposed to is the leaf surface temperature (in K), Fn is the net heat
direct solar radiation are relatively large, the expression of gain from radiation (in W/m2), C is the net sensible heat
thermal diffusion is essential for the accurate description of loss (in W/m2) and lE is the net latent heat loss (in W/m2).
the phenomenon. Radiative heat exchanges between the canyon surfaces
When air velocity is considered, eddy diffusion is much have been described analytically, according to the radiative
stronger than molecular diffusion (conduction) in the air in heat transfer theory [16] and not with the use of a
the atmosphere away from the boundary layer of the combined convection and radiative heat transfer coef-
surface. Despite the fact that molecular diffusion always cient. Thus the radiative heat exchanges between surfaces
takes place in the air, it is omitted from both heat and mass with different emissivities in closed enclosures is expressed
transfer in the air at these levels, as it is 104105 smaller by
than eddy diffusion [13]. The effect of vapour gradients N  
qi X 1
onto temperature in the air nodes well above the ground is   1 F ij qj H 0i
i j1
j
expressed through eddy diffusion coefcients [14,15]. Heat
and mass transfer in the two-dimensional model of the X
N

binary airwater vapour mixture thus becomes: F ij E bi  E bj for i 1; 2; . . . ; N, 8

j1
     
q q q q qT q qT
u w T K Hz K Hx , where qi is the radiation emitted from the surface i (W/m2),
qt qx qz qz qz qx qx ei is the i surfaces emmisivity and ej is the emmisivity of the
(3) rest of the surfaces, Fij is the view factor of surface i
      towards surface j, H0i is any external radiation arriving at
q q q q qq q qq surface i, and Ebi equals to sTi4, where s is the
u w q K Ez K Ex , (4)
qt qx qz qz qz qx qx StefanBoltzmann constant (5.67  108 W/m2 K4) and Ti
is the temperature of the ith surface.
where KHz and KEz are the eddy diffusion coefcient of Climatic characteristics, such as air temperature, relative
energy and water vapour, respectively, in the vertical axis humidity and wind speed, are set as the boundary nodes of
and KHx and KEx are the respective diffusion coefcients in the model, placed 10 m above the upper part of roofs.
x-axis. The expression of these diffusion coefcients is These climatic characteristics, as well as solar radiation
given by the MoninObukhov similarity theory [14,15]. derive from meteorological data from METEONORM
The water vapour gradients are taken into consideration in [17]. Solar radiation is input onto the surfaces, according to
the calculation of the eddy diffusion coefcients of energy. their orientation, inclination and shading pattern. The
Regarding solid materials, they are considered as a shading pattern, determined by the canyon geometry and
system, consisting of a capillary-porous building material the geographic latitude, was calculated with the software
in the medium of wet air and in a region of positive ECOTECT [18], where the same canyon geometries, as the
temperatures (no ice). The equations describing the one- ones described below were input, for the different latitudes
dimensional heat and mass transfer can be expressed and longitudes examined. Air velocities in the vicinity of
by [11] the canyon were calculated with the CFD code WinAir4
dT q2 T l qq [19]. WinAir4 is an in-house code, which uses the xed
ac 2 , (5) viscosity models, with a variation on simple solution
dt qz cc qt
scheme. The canyon geometries in the CFD model are
the same as for the heat and mass transfer model. As the
dq q2 q
am 2 , (6) CFD model is three-dimensional, the length of the canyon
dt qz was 40 m and the rest of the canyon dimensions (height,
where ac is the building material thermal diffusion width) varied, according to the canyon geometry, as
coefcient (in m2/s), cc is the building material specic discussed below. The CFD code mesh was also the same
heat capacity (in J/kg K), e is the evaporation number of as for the two-dimensional heat and mass transfer model,
the building material, and am is the diffusion coefcient of as described in Fig. 1. The grid is not uniform; near the
moisture in the building material (in m2/s). building and road surfaces the grid is 0.30 m, while two
 ARTICLE IN PRESS
E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493 483

nodes away from the surface, the grid varies according to
the canyon geometry, from 1 to 3 m for the canyon width,
and from 1 to 2 m for the canyon height. For the length, it
is constantly set at 2 m. At the boundary surface are set
10 m away from the roof of the buildings, 10 m away from
the windward surface and 100 m away from the leeward
surface when the building height is 5 m, and 10, 20 and
200 m, respectively, when the building height is 10 m. The
input climatic data of wind speed, air temperature and
relative humidity are input as the boundary conditions for
each hour of the diurnal prole. The heat gains on
buildings and the street are calculated according to the
solar radiation absorbed by the surface, which depends on
its orientation, and shading pattern, the latter having been
dened by [18]. The output air velocities from the CFD
code in the middle of the canyon are input at the respective
nodes of the two-dimensional heat and mass transfer
model.
Four types of vegetation covering the building envelope
are examined for each canyon geometry:
Buildings are made of concrete, and the street is covered
with asphalt. A summary of the hydrothermal properties of
the materials and vegetation considered in the canyons is
made in Table 1. All these cases are examined for nine cities
in nine different types of climates, where cities and
evapotranspiring vegetation can be found. Based on
Koeppens climatic classication [20], the nine cities
studied, and the climatic type in which they belong, are
summarised in Table 2. All cases are examined for a typical
day of their hottest month. Their climatic data have
derived from hourly data from METEONORM [17]. The
effect of vegetation on the urban texture of each city is
examined for its hottest month. For Athens, Hong Kong,
London, Montreal, Moscow and Riyadh, July is chosen as
the hottest month, while for Mumbai May is used, for
Beijing June, and for Braslia September. The typical day
Table 1
Hydrothermal properties of plants, soil, building materials (concrete) and
street materials

(a) a base case, where no green is placed in and around the Characteristic Concrete Asphalt Soil Plants
canyon, referred to as the no-green case, Specic thermal capacity 1.60 2.00 1.15 2.60
(b) the green-roofs case, where both roofs are covered (MJ/m3K)
with vegetation (ground-covering grasses) Thermal conductivity (W/mK) 1.70 1.30
(c) the green-walls case, where both walls inside the Vapour diffusivity (106 m2/s) 0.55 1.58
Ratio of vapour diffusion 0.20 0.10
canyon are covered with vegetation (ivies) and
coefcient to total moisture
(d) the green-all case, where both roofs and walls are diffusion coefcient
covered with vegetation. Emissivity 0.94 0.81 0.94 0.94
Albedo 0.23 0.10 0.23 0.30
Three types of canyon geometries are examined, accord- Hydraulic conductivity 0.01
(104 m/s)
ing to the wind ow developed in each:
Moisture potential, when soil is 49.0
saturated (cm)
(a) a canyon with height (H) 10 m and width (W) 5 m, Maximum volumetric water 0.492
referred to as H10W5 canyon, where, according to content (m3/m3)
Santamouris [6], skimming ow is developed, with very Coefcient b 10.40
Convective heat resistance 200
low air velocities, and sun shaded,
(s/m)
(b) a canyon with 5 m height and 10 m width, referred to as Resistance expressing the plant 100
H5W10 canyon, where wake interference ow is type (s/m)
developed, with bigger air velocities and more exposed Canopy extinction coefcient 1.4
surfaces to direct solar radiation and Level of soil moisture below 0.25
which permanent wilting of the
(c) the H5W15 canyon, with 5 m height and 15 m width,
plant occurs (m3/m3)
where isolated roughness ow is developed, with much
larger air velocities, and greater exposure to solar
radiation.
Table 2
The canyons are examined with two orientations: Table of cities studied

City Climate Location

(a) one where the canyons axis was parallel to the
EastWest axis (referred to as the EW canyon) and London, UK Temperate 51.32N, 0
Montreal, Canada Subarctic 45.31N, 73.34W
(b) one where the canyons axis was parallel to the
Moscow, Russia Continental cool summer 55.45N, 37.37E
NorthSouth axis (referred to as the NS canyon). Athens, Greece Mediterranean 37.59N, 23.43E
Beijing, China Steppe 39.48N, 116.23E
Two directions of wind ow are considered: Riyadh, Saudi Arabia Desert 24.38N, 46.43E
Hong Kong, China Humid subtropical 22.16N, 114.12E
Mumbai, India Rain forest 18.54N, 72.5E
(a) one aligned to (referred as x) and
Braslia, Brazil Savanna 15.48S, 47.54W
(b) one parallel to the canyons axis (referred as y).
 ARTICLE IN PRESS
484 E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493

Daytime
Maximum, minimum, average and day-time average values of climatic characteristics (total solar radiation on a horizontal plane, air temperature, relative humidity, wind speed) for the 24-h prole of
of the hottest month is a 24-h prole whose climatic data

average
are averaged for each hour of the month examined, to

5.2

4.0
4.2
4.2
4.6
4.5
3.8
5.2
4.0
make a diurnal typical climatic prole for the month
studied. The maximum, minimum, average and daytime

Average
average quantities of temperature, relative humidity, wind

4.8

3.8
3.8
4.2
4.3
4.3
3.3
4.8
3.8
speed and solar radiation on a horizontal plane are
presented in Table 3. In Figs. 25 the diurnal prole of
Min
Wind speed

3.5
3.3
3.2
3.0
3.9
3.3
3.5
2.3
3.8
air temperature, relative humidity, solar radiation on a
horizontal plane and wind speed are presented for the
Max

typical day of the hottest month of all nine cities.

6.4
4.3
4.7
4.9
4.4
5.5
5.2
4.6
5.9
Day-time

3. Discussion and analysis

average

45.9
57.9
52.1
70.9
70.5
62.1
71.6
62.4
32.4
3.1. Direct cooling effects
Average

Air and surface temperatures lower signicantly in all

50.2
62.7
58.6
77.1
74.3
66.1
74.7
69.5
36.2

climates examined, when walls and roofs are covered with

Relative humidity

vegetation, as can be observed in Figs. 710. The heat

Min

37.3
47.1
44.0
62.9
63.5
49.7
60.3
53.1
23.5

uxes on the vegetated and on the non-vegetated surfaces

are very different. As can be observed in Fig. 6 for a green
Max

roof and for a concrete roof in Montreal, the 24-h prole of

57.4
75.0
65.7
85.1
77.3
77.7
86.3
79.1
45.8

Total Solar Radiation on a Horizontal Plane

Daytime
average

900
Athens
27.0
25.6
25.4
30.9
18.0
22.3
18.0
30.8
38.3

800 Beijing
Solar Radiation (W/m2)

700 Brasilia
Hong Kong
Average

600 London
Montreal
25.7
24.2
23.5
29.5
17.2
21.2
17.3
28.9
36.5

500
Moscow
Air temperature

400 Mumbai
Riyadh
Min

22.3
19.6
20.4
26.7
14.5
17.1
13.9
25.3
31.2

300
200
Max

30.1
29.1
28.0
33.1
19.7
25.9
20.9
33.5
42.8

100
0
1 3 5 7 9 11 13 15 17 19 21 23
Time (Hours)
Daytime
average

Fig. 2. Twenty four-hour prole of the total solar radiation on a

331.9
457.5

437.8
417.2
286.5
376.5
297.9
502.1
466.9

horizontal plane, for the hottest month of each city, which is input on the
Total solar radiation on a horizontal plane

unshaded, horizontal surfaces of the heat and mass transfer model.

Average

Air Temperature
216.4
298.4

228.4
235.8
199.3
261.9
220.2
283.8
284.2

45
the typical day of the hottest month of each city

40
35
Temperature (C)

30
Min

0.0
0.0

0.0
0.0
0.0
0.0
0.0
0.0
0.0

25
20
15
589.2
809.0

726.0
704.3
497.3
692.2
527.8
846.9
850.0
Max

10 Athens Beijing Brasilia

5 Hong Kong London Montreal
Moscow Mumbai Riyadh
Braslia, September

0
Hong Kong, July

1 3 5 7 9 11 13 15 17 19 21 23
Montreal, July

Mumbai, May
Moscow, July
London, July
Beijing, June

Riyadh, July
Athens, July

Time (Hours)
Table 3

Fig. 3. Twenty four-hour prole of air temperature for the hottest month
City

of each city, which is input at the boundary nodes of the heat and mass
transfer model.
 ARTICLE IN PRESS
E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493 485

Relative Humidity external surface of the concrete roof is much larger than
100
the convective heat ux density in the upper part of the green
90
roofs canopy. For the concrete roof its 24-h prole ranges
80
Relative Humidity (%)

from 345.1 to 128.6 W/m2, while for the green roofs upper
70
surface, it only ranges from 51.3 to 99.9 W/m2. The
60
convective heat exchanges between the grass foliage and
50
the air are milder than those between the solid concrete
40
roof and the air. The total radiative heat ux density (both
30
short and long wave radiation) on the external surface of
20 Athens Beijing Brasilia
Hong Kong London Montreal
the concrete roof is also larger than that on the upper part
10
Moscow Mumbai Riyadh of the canopy layer. It ranges from 158.2 to 355.1 W/m2
0
1 3 5 7 9 11 13 15 17 19 21 23 on the concrete roof and from 38.8 to 229.5 W/m2 on the
Time (Hours) green roof. Due to the redistribution of radiation within
the vegetated layer, the total radiative heat exchanges are
Fig. 4. Twenty four-hour prole of relative humidity for the hottest smaller on the vegetated surface, when compared with the
month of each city, which is input at the boundary nodes of the heat and
mass transfer model.
concrete roof. As can be observed in Eq. (7), the conductive
heat component is omitted in the relationship governing
heat transfer in plants as too small [1315,21,22], while it is
Wind Speed an important factor in the heat transfer of a concrete roof,
7 with a range from 444.5 to 154.5 W/m2 on the external
part of the roof. Nonetheless, the greatest differences are
6
observed at the evaporative heat uxes, which range from
46.3 to 170.6 W/m2 for the concrete roof and from
Wind Speed (m/s)

4
593.2 to 26.4 W/m2 for the green roof. As the
evaporative heat transfer on the green roof acts constantly
3 as a heat sink and the radiative energy absorbed by the
2 green roof is smaller than that absorbed by the concrete
Athens Beijing Brasilia
roof, the energy uxes on a green surface can only offer
1 Hong Kong London Montreal lower surface and air temperatures, when compared to
Moscow Mumbai Riyadh
0 those produced by concrete surfaces.
1 3 5 7 9 11 13 15 17 19 21 23 Because of these energy distributions, canyon air
Time (Hours) temperature lowers the most when both walls and roofs
Fig. 5. Twenty four-hour prole of wind speed for the hottest month of
are covered with vegetation in all climates examined. This
each city, which is input at the boundary nodes of the heat and mass can be explained by the fact that when roofs are covered
transfer model. with vegetation, air masses enter the canyon much cooler,
from the vegetated roofs. On the other hand, when only
400 walls are covered with vegetation, air masses enter the
300
canyon heated by the plain roofs, which absorb the quite
200
Heat Flux Density (W/m2)

100 Decrease of canyon air temperature,

green-all case
0 0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
-100 4 8 12 16 20 24
Athens
-200 -2
Temperature Decrease (C)

Conv, rf-gr
Beijing
-300 Evap, rf-gr
-4 Brasilia
Rad, rf-gr
-400 Cond, rf-con HongKong
Conv, rf-con
-500 Evap, rf-con -6 London
Rad, rf-con
-600 Montreal
-8
Time (Hours) Moscow

-10 Mumbai
Fig. 6. Convective (Conv), evaporative (Evap), long and short-wave
radiative (Rad) and conductive (Cond) heat uxes on a concrete roof (rf- Riyadh
con) and on a green roof (rf-gr) in Montreal. -12
Time (Hours)
the convective, conductive, evaporative and radiative heat Fig. 7. Air canyon temperature decrease in the EW, H5W10 canyon, with
uxes on the green and the concrete roofs differ parallel wind ow, when both roofs and walls are covered with vegetation,
signicantly. The convective heat ux density at the for all climates examined.
 ARTICLE IN PRESS
486 E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493

great amounts of summer insolation. For this reason, in the Decrease of roof surface temperature
instance of Hong Kong, canyon air temperature decrease 5

Temperature Decrease (C)

Athens
reaches 8.4 1C maximum and 6.9 1C daytime average in the Beijing
green-all case (Fig. 7), while for the green-wall case these -5 4 8 12 16 20 24
Brasilia
numbers become only 3.9 and 2.5 1C, respectively (Fig. 8). HongKong
London
In general, the temperature decrease is quite signicant -15 Montreal
for both surface and air temperatures both inside the Moscow
canyon and at roof level. Regarding surface temperature -25 Mumbai
Riyadh
decrease of the south-oriented wall, it reaches from 18.7 1C
maximum and 14.3 1C daytime average for Riyadh to 9.8 -35
and 5.6 1C, respectively, for Moscow (Fig. 9). Roof surface Time (Hours)
temperatures lower even more, due to the greatest amounts Fig. 10. Roof surface temperature decrease when covered with vegetation,
of solar radiation horizontal surfaces receive in summer; for all climates examined.
the greatest day-time average temperature decrease is noted
for Riyadh (12.8 1C) and the greatest maximum for
Mumbai (26.1 1C), while the smallest decreases are noted Decrease of asphalt surface temperature
for Moscow and London (Fig. 10). Moscow reaches the 0.0
smallest daytime average surface temperature decrease 4 8 12 16 20 24

Temperature Decrease (C)

Athens
(9.1 1C), while London the smallest maximum (19.3 1C). In -0.5
Beijing
the subject of air temperature decrease inside the canyon Brasilia
-1.0 HongKong
for the green-all case, it reaches its peak for Riyadh London
(11.3 1C maximum and 9.1 1C daytime average), while its -1.5 Montreal
Moscow
Mumbai
-2.0
Decrease of canyon air temperature, Riyadh
green-wall case
0 -2.5
Time (Hours)
Temperature Decrease (C)

4 8 12 16 20 24
-1 Athens
Beijing Fig. 11. Asphalt temperature decrease when walls are vegetated in the
-2 Brasilia H5W10 canyon for all the nine climates examined.
HongKong
-3 London
Montreal
smallest decreases are noted in Moscow (3.6 and 3.0 1C,
-4 Moscow respectively) (Fig. 7). For the green-wall case, air tempera-
Mumbai ture decrease reaches its maximum again for Riyadh
-5 Riyadh (5.1 1C maximum and 3.4 1C daytime average) and its
-6 lowest decreases for Moscow (2.6 and 1.7 1C, respectively)
Time (Hours) (Fig. 8).
Fig. 8. Air canyon temperature decrease in the EW, H5W10 canyon, with
parallel wind ow, when only walls are covered with vegetation, for all 3.2. Indirect radiative cooling effects
climates examined.
On prima facie evidence, the air inside a canyon with
vegetated walls is reduced due to the evapotranspirational
Decrease of surface temperature of rate from plants and the lower surface temperatures of
south-oriented wall vegetated surfaces. The latter are responsible not only for
0
4 8 12 16 20 24 lowering the air temperature but also for lowering surface
temperatures of surfaces not covered with vegetation.
Temperature Decrease (C)

Athens
-5 Beijing As the radiative heat exchanges between the urban
Brasilia canyon surfaces have been modelled analytically, a
HongKong decrease is observed in the asphalt surface temperature
-10
London when walls are covered with vegetation. In Fig. 11 the
Montreal
decrease of the asphalt surface temperature is presented for
-15 Moscow
the H5W10 canyon for all the climates examined. As can be
Mumbai
Riyadh observed, the greatest decreases occur for hot and with
-20 high solar radiation Riyadh, with a maximum decrease of
Time (Hours) 2.0 1C and a daytime average 1.3 1C. The lowest surface
Fig. 9. Surface temperature decrease of the south-oriented wall, when
asphalt temperature decreases take place in much colder
covered with vegetation, in the EW, H5W10 canyon, for all climates and with lower insolation Moscow (maximum 0.9 1C,
examined. daytime average 0.6 1C). As the air temperature near the
 ARTICLE IN PRESS
E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493
ground is primarily affected by the horizontal surface
temperature [14,15,21], the effect of vertical surface
temperatures is not as crucial as that of the horizontal
surface. This radiative cooling of the street asphalt by green
walls, reaching up to 2.0 1C in hot Riyadh has an
additional effect on lowering air temperatures, apart from
evapotranspirational and convective cooling effects.
3.3. Climatic characteristics
It can be said with certainty that, the hotter and drier a
climate is, the more important the effect of green walls and
green roofs on mitigating urban temperatures is (Figs. 7
and 8). As can be observed in Figs. 3 and 4 and Table 3,
Riyadh is the hottest and most arid of all the cases
examined, with urban temperatures reaching 42.8 1C
maximum and 31.2 1C minimum, with a daytime average
of 38.3 1C, while relative humidity spans from 45.8% to
23.5%, with a daytime average of only 32.4%. These
extreme climatic inputs, benet the most from green walls
and green roofs and, as has been mentioned in Section 3.1,
the green-all case reaches temperature decreases of the
magnitude of 11.3 1C maximum and 9.1 1C daytime
average and the green-walls case 5.1 and 3.4 1C, respec-
tively. Much more humid Mumbai (Fig. 4) reaches smaller
decreases, of the magnitude of 6.6 1C daytime average and
8.0 1C maximum for the green-all case and 2.7 and 4.4 1C,
respectively, for the green-walls case. The colder climates
of London, Moscow and Montreal benet the least,
reaching daytime average decreases from 1.7 to 2.1 1C
and maxima from 2.6 to 3.2 1C for the green-walls case and
from 3.0 to 3.8 1C and from 3.6 to 4.5 1C, respectively, for
the green-all case.
3.4. Roof versus canyon
Temperatures at roof level decrease more than inside the
canyon, when the building envelope is covered with
vegetation. This is because the roof, being more exposed
to the much larger amounts of summer solar radiation on
the horizontal plane, raises its temperatures even more
when plain, low albedo-building materials are exposed to
direct solar gains. However, the canyon, due to its
geometry, is generally more shaded, not reaching the peak
temperatures roof surfaces do. By covering the roof with a
vegetated medium, which regulates its temperature so as
not exceed some crucial levels, roof temperatures decrease
more than the temperature inside the canyon, when both
roofs and walls are covered with vegetation. For all the
nine climates examined, the maximum temperature de-
crease at the air layer 1 m above the roof reaches from
26.0 1C for Riyadh to 15.5 1C for London and daytime
average temperatures from 12.8 1C for Riyadh to 5.8 1C for
Moscow (Fig. 12). The air inside the canyon reaches lower
decreases; for the green-all case the air temperature
decrease reaches a maximum from 11.3 1C for Riyadh to
3.6 1C for Moscow and daytime average from 9.1 1C for
487
Decrease of air temperature 1m above the roof
5
Temperature Decrease (C)
0 Athens
-5 4 8 12 16 20 24 Beijing
Brasilia
-10 HongKong
-15 London
Montreal
-20 Moscow
-25 Mumbai
Riyadh
-30
-35
Time (Hours)
Fig. 12. Air temperature decrease 1 m above the roof, for all climates
examined.
Riyadh to 3.0 1C for Moscow (Fig. 7). However, tempera-
tures at roof level start falling after 12:00, while for the
more stable conditions inside the canyon, temperatures due
to vegetation on walls start decreasing from early in the
morning, as can be observed by comparing Figs. 7 and 12.
3.5. Canyon orientation
Canyon orientation determines the shading pattern on
both the horizontal and the vertical parts of the canyon
geometry. It determines the amount of insolation received,
especially for the vertical planes, depending on their
orientation. During the summer months examined, the
amount of irradiation received on vertical planes is much
smaller than the horizontal one, for all orientations. Thus,
the orientation, despite the fact that it plays an important
role in temperature distributions in and around the canyon,
it does not affect temperature decreases so signicantly
when vegetation covers its vertical surfaces and roofs. The
magnitude of the effect strongly depends on the geographic
latitude. The examples of Hong Kong (22.16N) and Athens
(37.59N) are discussed below.
For all the climates examined, it has been observed that
the amount and geometry of vegetation is more important
than the canyons orientation. In the instance of Hong
Kong, solar radiation in all vertical orientations is not so
high, reaching a maximum of only 185 W/m2 for the
south orientation and 427 W/m2 for the west orientation
(Fig. 13). The green-walls case of the EW and NS oriented
H5W10 canyons result in 2.4 and 2.0 1C daytime average
temperature decrease, respectively, inside the canyon, with
an only 0.4 1C difference between the two orientations. For
the maximum, this difference becomes 0.7 1C (3.8 1C
maximum temperature decrease for the EW oriented
canyon and 3.1 1C for NS one). For the green-all case,
the differences between the two orientations become even
smaller, reaching 0.2 1C for the daytime average (tempera-
ture decrease being 6.8 1C for EW and 6.6 1C for NS), and
0.0 1C for the maximum (maximum temperature decrease
being 8.5 1C for both EW and NS orientation). It can be
observed in Fig. 14 that the amount and geometry of
 ARTICLE IN PRESS
488 E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493

Global rad, horiz. Global rad, horiz.

Globalrad, south Global rad, south
800 900
Global rad, north Global rad, north
700 800 Global rad, east
Global rad, east
Global rad, west
Solar Radiation (W/m2)

600 Global rad, west 700

Solar Radiation (W/m2)

500 600
500
400
400
300
300
200
200
100
100
0 0
1.00 4.00 7.00 10.00 13.00 16.00 19.00 22.00 1 4 7 10 13 16 19 22
Time (Hours) Time (Hours)

Fig. 13. Hong Kong global solar radiation on horizontal and vertical Fig. 15. Athens global solar radiation on horizontal and vertical planes of
planes of east, west, south and north orientation in July. east, west, south and north orientation in July.

35 0 Ta,can[no gr]EW-HgKg 40 0 Ta,can[no gr]Ath-EW

Temperature Decrease (C)

Temperature Decrease (C)

30 -1 35 -1
Ta,can [no gr]NS-HgKg Ta,can[no gr]Ath-NS
-2
Temperature (C)

30 -2
Temperature (C)

25 Ta,can[gr a]EW-HgKg Ta,can[gr a]Ath-EW

-3
Ta,can[gr a]NS-HgKg 25 -3 Ta,can[gr a]Ath-NS
20 -4
Ta,can[gr w]EW-HgKg 20 -4 Ta,can[gr w]Ath-EW
15 -5
Ta,can[gr w]NS-HgKg 15 -5 Ta,can[gr w]Ath-NS
-6
10 10 -6 DTa,can[gr a]Ath-EW
-7 DTa,can[gr a]EW-HgKg
5 -8 DTa,can[gr a]NS-HgKg 5 -7 DTa,can[gr a]Ath-NS
0 -9 DTa,can[gr w]EW-HgKg 0 -8 DTa,can[gr w]Ath-EW
12 15 18 21 DTa,can[gr w]NS-HgKg 12 15 18 21 DTa,can[gr w]Ath-NS
Time (Hours) Time (Hours)

Fig. 14. Hong Kong temperature distributions and decreases inside the Fig. 16. Athens temperature distributions and decreases inside the canyon
canyon for no-green [no-gr], green-all [gr-a], green-walls [gr-w], for EW for no-green [no-gr], green-all [gr-a], green-walls [gr-w], for EW- and NS-
and NS oriented H5W10 canyon. oriented H5W10 canyon.

vegetation mostly affect the temperature decrease and not being 6.6 1C for EW and 6.8 1C for NS orientation). For
so much the canyons orientation. The green-all case in the the green-walls case these differences become larger,
EW-oriented canyon results in 5.4 1C higher for the reaching 0.8 1C for the daytime average (temperature
daytime average and 4.7 1C higher for the maximum decrease being 3.0 1C for EW and 2.2 1C for NS) and
temperature decrease than the green-walls case. For the 1.2 1C for the maximum (temperature decrease being 4.5 1C
NS orientation, these differences become 4.6 and 4.4 1C, for EW and 3.3 1C for NS). Yet again, the difference
respectively. between the two amounts of vegetation (green-all and
However, in the example of Athens, the amount of green-walls) is more crucial than the difference between the
irradiation received in the east and west oriented vertical decreases of different orientations. The difference between
planes is much larger than on the south and north the temperature decrease of the green-all and the green-
orientations and proportionally larger to those of Hong walls cases of the EW oriented canyon reaches 2.6 1C for
Kongs. The maximum solar radiation received by the the daytime average and 2.1 1C for the maximum. For the
south-oriented vertical plane is 374.0 W/m2, while for the NS orientation, these differences become larger, 3.2 and
east-oriented plane reaches the magnitude of 616.7 W/m2 3.4 1C, respectively (Fig. 16).
(Fig. 15). This has a direct effect on the way the canyon In general, it can be concluded that the orientation may
orientation affects temperature decreases due to vegetated play a countable role in temperature decreases due to
surfaces. For the green-all case, the difference between the vegetation, only when the amounts of solar radiation
temperature decrease of the air inside the canyon remains received by the vertical planes differ signicantly. Yet
small, of the magnitude of 0.1 1C for the daytime average again, concerning temperature decreases, the amount of
(temperature decrease being 5.6 1C for EW and 5.5 1C for vegetation placed on buildings is more crucial than the
NS), and 0.2 1C for the maximum (temperature decrease orientation of the canyon, with the green-all case, when
 ARTICLE IN PRESS
E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493
both roofs and walls are covered with vegetation, leading
to much larger temperature decreases.
3.6. Canyon geometry
For all geometries and cases examined, it can be
concluded that the wider a canyon is the smaller the effect
of green roofs and green walls on its temperature decreases.
For wider canyons, temperatures inside the canyon are
dominated by the proportionally larger street surface and
the fact that it is more exposed to direct solar radiation. In
Riyadh, temperature decreases in the wide H5W15 canyon
are of the magnitude of 1.2 1C for the daytime average and
1.7 1C for the maximum of the green-wall case and 7.3 and
9.3 1C, respectively, for the green-all case. For the narrower
H10W5 canyon these decreases reach 6.3 1C daytime
average and 9.1 1C maximum for the green-walls case and
8.9 and 12.3 1C, respectively, for the green-all case (Figs. 17
and 18). It can be observed in Fig. 18 that the wider a
canyon is the smaller the effect of green walls is on its
temperatures. Nevertheless, when the combination of green
roofs and green walls is implemented (green-all case),
temperature decreases rise signicantly. In contrast, for the
narrow H10W5 canyon, whose walls are proportionally
more dominant than the street, the green-walls case has a
Air temperature decrease for different canyon
geometries in Riyadh for the green-all case
0 differences between the temperature decreases of the two
Temperature Decrease (C)
7 11 15
-5
-10
DTa [gr a]H5W10
DTa [gr a]H10W5
DTa [gr a]H5W15
-15
Time (Hours)
Fig. 17. Air temperature decrease during the day in the H5W10, H10W5
and H5W15 canyon, for the green-all case, Riyadh.
Air temperature decrease for different canyon
geometries in Riyadh for the green-wall case
0 -3
7 11 15
Temperature Decrease (C)
-2
-4
-6
DTa[gr w]H5W10
-8
DTa[gr w]H10W5
DTa[gr w]H5W15
-10
Time (Hours)
Fig. 18. Air temperature decrease during the day in the H5W10, H10W5
and H5W15 canyon, for the green-wall case, Riyadh.
489
signicant effect on lowering urban temperatures. The
combination of both green roofs and green walls does not
lead to such signicant further decreases, as is the case in
the wider H5W10 and H5W15 canyons. The differences
between the temperature decreases of the green-all and the
green-walls case reaches an average of 5.7 and 6.1 1C,
respectively, for the H5W10 and the H5W15 canyons,
while for the H10W5 canyon it is only 2.6 1C. For the
maximum, the discrepancy between the two wider and the
narrower canyon is even larger, reaching 6.2, 7.6 and
3.2 1C, respectively.
3.7. Wind direction
Wind direction affects temperature decreases inside the
canyon even less than orientation. Although it is a
signicant factor for temperature distributions, for the
decreases due to vegetation it is the vegetation itself that
plays the most important role. The differences between
temperature decreases in the same canyons for different
wind directions are insignicant. In Fig. 19 temperature
distributions and temperature decreases for the EW-
oriented H5W10 canyon in Mumbai are presented, for
both parallel (y) and perpendicular (x) to the canyons
axis wind directions. It can be observed, that for the low air
velocities inside the canyon, temperature distributions are
not so different for the two wind directions as they were for
the canyon orientations (Figs. 14 and 16). Temperature
19
wind directions become quite insignicant, reaching a
0.1 1C difference for both green-walls and green-all cases. It
can also be observed that for both wind directions the most
important factor for temperature decreases is the amount
of vegetation, with the green-all case reaching temperature
decreases 3.7 1C higher than the green-walls case.
For the wider H5W15 canyon, with its much larger air
velocities, temperature decreases are similar for the two
wind directions (Fig. 20). The differences between the
temperature decreases of the two wind directions reach a
maximum of 0.3 1C, with 0.2 1C daytime average for the
40 0 Ta,can[no gr]Mumb-EW-x
Temperature Decrease (C)
35 -1 Ta,can[no gr]-Mumb-EW-y
-2
Temperature (C)
30 Ta,can[gr a]Mumb-EW-x
25 Ta,can[gr a]-Mumb-EW-y
19
-4 Ta,can[gr w]Mumb-EW-x
20
-5 Ta,can[gr w]-Mumb-EW-y
15
-6 DTa,can[gr a]Mumb-EW-x
10 -7 DTa,can[gr a]-Mumb-EW-y
5 -8
DTa,can[gr w]Mumb-EW-x
0 -9 DTa,can[gr w]-Mumb-EW-y
12 15 18 21
Time (Hours)
Fig. 19. Mumbai temperature distributions and decreases inside the
canyon for no-green [no-gr], green-all [gr-a], green-walls [gr-w], for a
parallel (y) and perpendicular (x) to the canyons axis wind direction in
the EW-oriented H5W10 canyon.
 ARTICLE IN PRESS
490 E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493

45 1 Ta,can[no gr]MumbH5W15-EW-x

Temperature Decrease (C)

40 0 Ta,can[no gr]-MumbH5W15-EW-y
Temperature (C)

35 -1
Ta,can[gr a]MumbH5W15-EW-x
30 -2
25 -3 Ta,can[gr a]-MumbH5W15-EW-y
20 -4 Ta,can[gr w]MumbH5W15-EW-x
15 -5
Ta,can[gr w]-MumbH5W15-EW-y
10 -6
5 -7 DTa,can[gr a]MumbH5W15-EW-x

0 -8 DTa,can[gr a]-MumbH5W15-EW-y
12 15 18 21
Time (Hours)

Fig. 20. Mumbai temperature distributions and decreases inside the

canyon for no-green [no-gr], green-all [gr-a], green-walls [gr-w], for a
parallel (y) and perpendicular (x) to the canyons axis wind direction in
the EW-oriented H5W15 canyon.
Fig. 22. PET for the EW-oriented H5W10 canyon, for the no-green [no
gr] and green-all [gr-a] cases, inside the canyon (EW) and on the roof (rf),
for Athens.

Fig. 21. PET for the EW-oriented H5W10 canyon, for the no-green [no
gr] and green-all [gr-a] cases, inside the canyon (EW) and on the roof (rf),
for Moscow.
Fig. 23. PET for the EW-oriented H5W10 canyon, for the no-green [no
green-wall case and 0.4 and 0.3 1C, respectively, for the gr] and green-all [gr-a] cases, inside the canyon (EW) and on the roof (rf),
green-all case. Again, the amount of vegetation has a for Riyadh.
stronger effect than the wind ow direction, even in the
wider canyon. It can be observed in Fig. 21, that for the much milder
It can thus be concluded, that for the generally low air summer of Moscow, the greening of the building envelope
velocities inside the urban canyons [6], the effect of wind does not lead to such major improvements of the outdoors
direction is not so strong on temperature decreases due to thermal comfort. PET ranging from slightly warm and
vegetated roofs and walls, as is the amount and geometry comfortable levels on the roof and inside the canyon for
of vegetation itself. the no-green case, lowers to cooler levels, from comfortable
to slightly cool, during daytime, when roofs and walls are
covered with vegetation (green-all case). Although moving
4. Thermal comfort from slightly warm to comfortable might not be so
spectacular, it could prove to be benecial for the thermal
In order to assess the thermal comfort improvements in comfort and well being of populations used to cooler
outdoors spaces when walls and roofs are covered with climatic conditions.
vegetation, the physiological equivalent temperature (PET) For much hotter Athens (Fig. 22) and Riyadh (Fig. 23),
is used, its expression deriving from Ref. [23] and its the improvements of outdoors thermal comfort are more
relationship with thermal sense from Ref. [24]. The results dramatic. For both climates, the bare concrete roof reaches
for the EW-oriented H5W10 green-all and no-green cases the very hot level in the afternoon. When covered with
are presented here, for Moscow, Athens and Riyadh in vegetation, the sensation warm is reached only for 4 h in
Figs. 2123. Emphasis is given on thermal comfort, not Athens and 5 in Riyadh. Most of the daytime, the exposed
only inside the canyon (symbolised with EW in the graphs), to direct solar radiation roof reaches the slightly warm
but also at the roof level (symbolised with rf). and comfortable zone for both cities. For inside the
 ARTICLE IN PRESS
E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493
canyon, the thermal sensation improves even more, from
hot, in both cases to slightly warm and comforta-
ble, reaching even slightly cool for both cases in the
early morning and late evening hours and even cool for
Athens, in the early morning hours.
In general, green roofs and walls can improve outdoors
thermal conditions not only at street level, but also at roof
level, turning these empty urban spaces into potentially
usable ones, in the form of superterrestrial gardens. By
covering roofs and walls with vegetation, thermal comfort
in the built environment can improve signicantly, not only
for hot climates, but for cooler ones, in which populations
are acclimatised to lower temperatures.
5. Energy savings from green walls and roofs
Apart from creating outdoor conditions, which are more
human-friendly, from a thermal point of view, green
roofs and green walls can also prove benecial for indoor
thermal conditions. In addition to the fact that they add a
further insulation layer to the buildings fabric, they can
decrease cooling load demands inside the building quite
signicantly due to the microclimatic modications dis-
cussed in this paper.
In a simplied steady-state analysis, without taking into
consideration internal thermal gains, heat gains/losses (qE)
from the buildings fabric with an average U-value U, an
indoors temperature Tin and an outdoors temperature Tout
are given by the relationship:
qE UT out  T in . (9)
For the no-green base case [no gr], the cooling load for
the non-vegetated canyon is given by the relationship:
 
qEno gr U T no gr  T in , (10)
where T[no gr] is the averaged air temperature inside the
canyon when no vegetation is placed either on walls or on
roofs. For the green-all case, with an average air
temperature inside the canyon T[gr a], heat gains are1:
 
qEgr a U T gr a  T in . (11)
Thus, the decrease in the cooling load, when both walls
and roofs are covered with green is given by
qno gr  qgr a
DqEgr a ),
qno gr
T no gr  T gr a
DqEgr a . 12
T no gr  T in
For T[no gr]6Tin, T[no gr]4Tin and T[gr a]4Tin.
1
Despite the fact that the U-value is altered, when vegetation is placed
on the buildings fabric, leading to further cooling load decreases, this is
not taken into consideration here, as the aim is to directly compare
between the effects of the microclimatic alterations on the buildings
cooling load, without it being affected by alterations to the fabric.
491
Fig. 24. Average cooling load decreases (%), with a 23 1C indoors
temperature, for the green-all and green-walls cases of all the climates
examined.
Similarly, for the green-walls case, if T[gr w] is the
average air temperature inside the canyon with the green
walls, the cooling load decrease becomes:
T no gr  T gr w
DqEgr w . (13)
T no gr  T in
For T[no gr]6Tin, T[no gr]4Tin and T[gr w]4Tin.
Considering an indoor limit temperature for cooling of
23 1C for all climates studied, the cooling load decreases
due to green-all and green-walls cases are given as a
daytime average in Fig. 24 and for an hourly basis in
Fig. 25.
As can be observed in Fig. 24, the largest cooling load
decreases in all climates examined, occur for the green-all
case. For the geometries examined for Braslia and Hong
Kong, the cooling load decreases for the green-all case
reach 100%; no cooling load is needed after covering roofs
and walls with vegetation, while in both cities cooling load
is needed in the afternoon and early evening hours for the
no-green case (Fig. 25c and d). London and Moscow are
not affected, regarding cooling loads, as no cooling load is
needed for the typical day examined, even before vegeta-
tion was placed around the canyon. Riyadh experiences a
quite high cooling load decrease, of the magnitude of 90%,
as does Montreal (85%) for the green-all case, lowering
their total hours of cooling demand from 12 to 5 and from
8 to 4, respectively (Fig. 25e and g). For the green-all case,
Mumbai reaches a 72% decrease, lowering its cooling
energy demand from 11 h to 6 (Fig. 25f), while for Athens
and Beijing the decrease is 66% and 64%, respectively,
lowering their energy demand by 4 and 3 h, respectively
(Fig. 25a and b).
For the green-walls cases, cooling load decreases are less
dramatic. The largest one is noted for Braslia (68%), with
6 h decrease in cooling demand (Fig. 25d). It is followed by
a 66% and 2 h decrease for Hong Kong (Fig. 25e), 52%
and 2 h for Montreal (Fig. 25e), 43% and 2 h for Athens
(Fig. 25a), 37% and 2 h for Beijing (Fig. 25b), 37% and 3 h
for Riyadh (Fig. 25(g) and 35% and 3 h for Mumbai
(Fig. 25f). It can be noted that the differences between the
green-all and green-walls cooling loads are smaller for
 ARTICLE IN PRESS
492 E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493

Fig. 25. Cooling load decreases (%) for (a) Athens, (b) Beijing, (c) Hong Kong, (d) Braslia, (e) Montreal, (f) Mumbai and (g) Riyadh for green-all and
green-walls cases.

humid climates (of the magnitude of 3237%) and greater
for arid climates (53% for Riyadh), due to the different
humidity concentrations in the two climatic groups.
In general, green roofs and green walls cool the
microclimate around them, which can lead to quite
important energy savings for cooling, depending on the
climatic type, the amount and position of vegetation on the
building. In cases where little cooling load is needed,
cooling demand can be reduced to zero by covering
building surfaces with vegetation. In other cases, energy
savings can also be signicant, varying from 90% to 35%.2
2 reach up to 26.0 1C maximum and 12.8 1C day-time
These percentages can become even greater, when a higher than 23 1C
limit temperature for cooling is considered. In general, inhabitants of hot
climates are accustomed to higher temperatures (in the instance of Greek
regulations, the limit temperature for cooling is set to 26 1C).
In addition to the energy savings themselves, this could
lead to successful applications of further passive cooling
techniques, especially ones employing ventilation, which
are not easy to implement in the extremely hot urban
conditions, in cases of large heat island densities.
6. Conclusions
From this quantitative research, it has been shown that
there is an important potential of lowering urban
temperatures when the building envelope is covered with
vegetation. Air temperature decreases at roof level can
average (Riyadh), while inside the canyon decreases reach
up to 11.3 1C maximum and 9.1 1C daytime average, again
for hot and arid Riyadh. It can be concluded that the
 ARTICLE IN PRESS
E. Alexandri, P. Jones / Building and Environment 43 (2008) 480493
hotter and drier a climate is, the greater the effect of
vegetation on urban temperatures. However, it has been
pointed out that also humid climates can benet from
green surfaces, especially when both walls and roofs are
covered with vegetation, reaching up to 8.4 1C maximum
temperature decrease for humid Hong Kong. Temperature
decrease due to vegetation is primarily affected by the
vegetation itself (amount and geometry), more than the
canyon orientation in hot periods. In general, the larger
amounts of solar radiation a surface receives, the larger its
temperature decreases are when it is covered with vegeta-
tion. For the low air velocities inside the canyon, the wind
direction does not have any signicant effect on tempera-
ture decreases due to vegetation.
Regarding the urban geometry, the wider a canyon is,
the weaker the effect green roofs and green walls have on
temperature decrease. For all climates examined, green
walls have a stronger effect than green roofs inside the
canyon. Nonetheless, green roofs have a greater effect at
roof level and, consequently, at the urban scale. The
combination of both green roofs and green walls leads to
the highest mitigation of temperatures inside the canyon. If
applied to only one unit block, green roofs and green walls
can create a small area of mitigated temperatures to the
urban heat island effect, as has been shown in this
microclimatic study. If applied to the whole city scale,
they could mitigate raised urban temperatures, and,
especially for hot climates, bring temperatures down to
more human-friendly levels and achieve energy saving
for cooling buildings from 32% to 100%.
Acknowledgements
This research has been funded by the State Scholarship
Foundation of Greece (IKY) from 2001 to 2003. The
authors are extremely grateful to Panagiotis Doussis for his
guidance and contribution to computer modelling.
References
[1] Benevolo L. The history of the city. London: Scholar Press; 1980.
[2] Koppe C, Kovats S, Jendritzky G, Menne B. Health and global
environmental change; heat-waves: risks and responses, series no. 2,
energy, environment and sustainable development. Copenhagen:
World Health Organization; 2004.
[3] White KS, et al. Technical summary, climate change 2001: impacts,
adaptation and vulnerability. In: McCarthy JJ, Canziani OF, Leary
NA, Dokken DJ, White KS, editors. Climate change 2001: impacts,
493
adaptation and vulnerability. Cambridge: United Nations Inter-
governmental Panel on Climate Change (IPCC) Cambridge Uni-
versity Press; 2001. p. 1973.
[4] Daz J, Ballester F, Lopez-Velez R. Impacts on human health. In:
Moreno Rodrguez JM, editor. The preliminary assessment of the
impacts in Spain due to effects of climate change, project ECCE.
Madrid: Ministry of the Environment; 2005.
[5] Davies RE, Knappenberger PC, Novicoff WM, Michaels PJ. Decadal
changes in summer mortality in US cities. International Journal of
Biometeorology 2003;47:16675.
[6] Santamouris M, editor. Energy and climate in the urban built
environment. London: James & James; 2001.
[7] Dimoudi A, Nikolopoulou M. Vegetation in the urban environment:
microclimatic analysis and benets, PRECis project, the European
Commission, Directorate General XII, Joule III, contract JOR3-
CT97-0192. Centre for Renewable Energy Sources, Pikermion,
2000.
[8] Bruse M, Fleer H. Simulating surfaceplantair interactions inside
urban environments with a three dimensional numerical model.
Environmental Modelling and Software 1998;13:37384.
[9] Giridharam R, Ganasan S, Lau SSY. Daytime urban heat island in
high-rise and high-density residential developments in Hong Kong.
Energy and Buildings 2004;36:52534.
[10] Versteeg HK, Malalasekera W. An introduction to computational
uid dynamics. Harlow, London: Prentice-Hall; 1995.
[11] Luikov AV. Heat and mass transfer in capillary-porous bodies.
Oxford: Pergamon Press; 1961.
[12] Eckert ERG, Drake RM. Heat and mass transfer. New York:
McGraw-Hill Book Company Inc.; 1959.
[13] Noilhan J. Les facteurs physiques du microclimat au voisinage dun
batiment. Etude bibliographique. Centre Scientique et Tecnique du
Batiment, Etablissement de Nantes, 1979.
[14] Jacobson MZ. Fundamentals of atmospheric modeling. Cambridge:
Cambridge University Press; 1999.
[15] Pielke RA. Mesoscale meteorological modeling. San Diego: Aca-
demic Press; 2002.
[16] Modest MF. Radiative heat transfer. New York: McGraw-Hill
International Editions; 1993.
[17] Remund J, Lang R, Kunz S, Meteonorm, version 3.0, Meteotest,
Bern, 1997.
[18] Marsh A. Ecotect, version 5.20(b), square one research PTY LTD,
2003.
[19] Welsh School of Architecture, Cardiff University, WinAir4. Cardiff,
2003.
[20] Critcheld HJ. General climatology. London: Prentice-Hall Inter-
natioal Inc.; 1974.
[21] Geiger R. The climate near the ground. Cambridge, Massachusetts:
Harvard University Press; 1965.
[22] Monteith JL, Unsworth M. Principles of environmental physics.
London, New York: Edward Arnold; 1990.
[23] Matzarakis A, Mayer H, Rutz F. Radiation and thermal comfort.
Sixth Hellenic conference in meteorology, climatology and atmo-
spheric physics, Ioannina 2; 2002. p. 73944.
[24] Kuttler W. Stadklima, Teil 2: phanomene und wirkungen. In: Moller
D. (Ed.), Beitragsserie klimaanderung und klimaschutz, UWSF,
vol. 16(4); 2004. p. 26374.