This article was downloaded by: [Massachusetts Institute of Technology], [Bruno Bueno]

On: 31 August 2012, At: 12:26

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Journal of Building Performance Simulation

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tbps20

The urban weather generator

a a b b
Bruno Bueno , Leslie Norford , Julia Hidalgo & Grégoire Pigeon
a
Massachusetts Institute of Technology, Cambridge, USA
b
CNRM-GAME, Météo-France and CNRS, Toulouse, France

Version of record first published: 31 Aug 2012

To cite this article: Bruno Bueno, Leslie Norford, Julia Hidalgo & Grégoire Pigeon (2012): The urban weather generator,
Journal of Building Performance Simulation, DOI:10.1080/19401493.2012.718797

To link to this article: http://dx.doi.org/10.1080/19401493.2012.718797

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
 Journal of Building Performance Simulation
2012, 1–13, iFirst article

The urban weather generator

Bruno Buenoa*, Leslie Norforda, Julia Hidalgob and Grégoire Pigeonb
a
Massachusetts Institute of Technology, Cambridge, USA; bCNRM-GAME, Me´te´o-France and CNRS, Toulouse, France
(Received 15 May 2012; final version received 20 July 2012)
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012

The increase in air temperature produced by urbanization, a phenomenon known as the urban heat island (UHI)
effect, is often neglected in current building energy simulation practices. The UHI effect can have an impact on the
energy consumption of buildings, especially those with low internal heat gains or with an inherent close interaction
with the outdoor environment (e.g. naturally-ventilated buildings). This paper presents an urban weather generator
(UWG) to calculate air temperatures inside urban canyons from measurements at an operational weather station
located in an open area outside a city. The model can be used alone or integrated into existing programmes in order
to account for the UHI effect in building energy simulations. The UWG is evaluated against field data from Basel
(Switzerland) and Toulouse (France). The error of UWG predictions stays within the range of air temperature
variability observed in different locations of the same urban area.
Keywords: urban heat island; urban climate; building energy model; urban canopy model; weather data

1. Introduction boundary layer, and its canopy-scale component due

Urban areas are affected by an increase in air
temperatures relative to the surrounding rural areas,
a phenomenon known as the urban heat island
(UHI) effect. The UHI effect has been documented
based on meteorological experiments carried out in
different cities around the world (Roth 2007, Hicks
et al. 2010, Lee and Baik 2010, Zhou and Shepherd
2010, Houet and Pigeon 2011). This phenomenon
has a characteristic diurnal pattern (Oke 1987), being
more intense at late afternoon and night. In the
morning, the urban–rural temperature difference can
be negative, what is called the urban cool island
(UCI) effect.
The observed UHI effect can be explained by the
different morphology of the urban terrain relative to
the rural terrain. The effective albedo tends to be
higher due to the inter-reflections between urban
surfaces. Additionally, the urban terrain delays the
diurnal cycle of air temperature due to the fact that
there is more surface exposed to the environment,
increasing the effective thermal inertia (Erell and
Williamson 2007). Furthermore, urban surface rough-
ness decreases the mean wind velocity and reduces the
convective heat removal. Added to this is the heat gain
due to anthropogenic sources (Sailor 2011) and the
lower evaporation due to the reduction of vegetated
areas. To analyse the UHI effect, it is useful to
differentiate between its mesoscale component, pro-
duced by the aggregated effect of the city on the urban
to the urban canyon effect (Kuttler 2008).
Building energy simulation programmes use stan-
dard meteorological databases obtained from measure-
ments at operational weather stations for annual
energy calculations. Operational weather stations are
usually located in open areas, without nearby obstruc-
tions and outside the city, typically at the airport.
Therefore, air temperature measurements might not
include the UHI effect.
The UHI effect can affect the energy performance of
buildings. Crawley (2008) studies the impact of the UHI
effect on a small office building by assuming typical
diurnal cycles of urban–rural temperature difference. His
study suggests that the UHI effect modifies the energy
consumption of office buildings between 5% and 10%
(increasing their cooling energy demand in summer and
decreasing their heating energy demand in winter).
Bueno et al. (2012c) predicts a similar impact of the
UHI effect on commercial buildings (i.e. office buildings)
and concludes that these buildings are less affected than
residential buildings by this phenomenon because their
energy performance is usually dominated by internal
heat gains. The study shows that the energy demand of
residential buildings can be modified by 20% for a
typical 4 K daily-maximum UHI effect.
At the same time as the UHI effect modifies the
indoor energy operation, the energy performance of
buildings can have an impact on outdoor air tempera-
tures, mainly through the waste heat emissions from

*Corresponding author. Email: bbueno@mit.edu

ISSN 1940-1493 print/ISSN 1940-1507 online

Ó 2012 International Building Performance Simulation Association (IBPSA)
http://dx.doi.org/10.1080/19401493.2012.718797
http://www.tandfonline.com
 2 B. Bueno et al.
outdoor air-conditioning equipment. Consequently,
there are situations in which the interactions between
the indoor and outdoor environments are reciprocal and
thus both domains have to be solved simultaneously.
This paper proposes an urban weather generator
(UWG) to calculate urban air temperatures using
meteorological information measured at an opera-
tional weather station and accounting for the recipro-
cal interactions between building and the urban
climate. Other studies that calculate urban weather
information through meteorological modelling can be
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012
found in the literature. Erell and Williamson (2006)
presented a rural-to-urban weather transformation
(the CAT model) based on the urban canopy model
LUMPS (Local-Scale Urban Meteorological Parame-
terization Scheme; Grimmond and Oke 2002) and the
CTTC (Cluster Thermal Time Constant) model (Swaid
and Hoffman 1990), which requires the calibration of
empirical parameters at the location of analysis and
neglects the UHI effect at mesoscale level. Oxizidis
et al. (2008) proposed a method for generating urban
weather files by coupling EnergyPlus (Crawley et al.
2001) with computational fluid dynamics and mesos-
cale atmospheric simulations. Kikegawa et al. (2003)
coupled a building energy model, an urban canopy
model and a mesoscale atmospheric model. The UWG
is presented as an alternative to computationally
expensive simulations. Its computational cost is
intentionally kept at the same order of magnitude as
annual building energy simulations.
This paper first describes the physics behind the
UWG. Then, the model is evaluated against field data
from two boundary-layer experiments, one carried out
in Basel, Switzerland (Rotach et al. 2005) and another
one carried out in Toulouse, France (Masson et al.
2008). A discussion of the limitations and prospects of
the UWG is presented at the end.
2. Model description
The UWG calculates hourly values of urban air
temperature and humidity given the weather data
measured at an operational weather station located
outside a city. The model is composed of four coupled
modules (Figure 1): the Rural Station Model (RSM),
which calculates sensible heat fluxes at the weather
station; the Vertical Diffusion Model (VDM), which
calculates vertical profiles of air temperature above the
rural site; the Urban Boundary-Layer (UBL) model,
which calculates air temperatures above the urban
canopy layer (above urban canyons) and the Urban
Canopy and Building Energy Model (UC-BEM), which
calculates urban sensible heat fluxes and urban canyon
air temperature and humidity. The interrelations between
modules are illustrated in Figure 2. The atmospheric
component of the UWG, composed of the VDM and
UBL model, is described in detail in Bueno et al. (2012a).
2.1. Rural Station Model
The RSM is a rural canopy model that reads hourly
values of meteorological fields measured at the rural
site and calculates sensible heat fluxes, which are then
provided to the VDM and the UBL model.
The model is based on an energy balance at the soil
surface. A transient heat diffusion equation represents
the storage and release of heat from the ground.
Dividing the soil in discrete layers, the RSM solves the
following system of equations:
@T1
d1 ðrcÞ1 ¼ C1;2 ðT2  T1 Þ þ Qsurf ð1Þ
@t
for the first layer,
@Ti
di ðrcÞi ¼ Ci;iþ1 ðTiþ1  Ti Þ þ Ci;i1 ðTi1  Ti Þ ð2Þ
@t
for each intermediate layer and
@Tn1
dn1 ðrcÞn1 ¼ Cni;n ðTdeep  Tn1 Þ ð3Þ
@t
for the deepest layer. In Equations (1)–(3), di, (rc)i and
Ti represent the depth, the volumetric heat capacity [J
m73 K71] and the average temperature of the layer i,
respectively; Ci,j is the mean thermal conductance over
the distance between two layers [W m72 K71]; Qsurf is
the sum of net-radiation, sensible and latent heat fluxes
at the surface and Tdeep is the annual-average air
temperature of the site, used as boundary condition
deep into the ground.
Radiation heat fluxes are measured at the weath-
er station. Surface sensible heat fluxes are computed
by using convective heat transfer coefficients (CHTC)
(see next section). Latent heat fluxes due to the
evapotranspiration of vegetation (if present) are
calculated as a fraction of the absorbed shortwave
radiation. More sophisticated vegetation models,
usually applied to numerical atmospheric models,
can be found in the literature (Noilhan and Mahfouf
1996). However, they contain various empirical
parameters for which it is difficult to find values if
detailed information about the soil and plant
composition is not available.
A suitable model for the accuracy requirements of
the UWG takes advantage of the fact that the ratio
between the latent heat flux and the net radiation is
 Journal of Building Performance Simulation 3
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012

Figure 1. Diagram of the urban weather generator (UWG) scheme, which is composed of four modules: the Rural Station
Model (RSM), the Vertical Diffusion Model (VDM), the Urban Boundary-Layer (UBL) model and the Urban Canopy and
Building Energy Model (UC-BEM). Thermal networks indicate the main heat transfer processes included in the RSM and UC-
BEM. Trur, Tubl and Turb represent the air temperature measured at the weather station, calculated at the urban boundary layer
and calculated at the urban site. The RSM provides rural sensible heat fluxes (Hrur) to the VDM and the UBL model. The UC-
BEM provides urban sensible heat fluxes (Hurb) to the UBL model.

Figure 2. Information exchanged among the different modules of the UWG.

relatively constant during daytime. The net radiation is typically presents small variations during the day. In
the sum of the net shortwave radiation (that which is the absence of light, plants’ stomata are usually closed
absorbed) and the net longwave radiation, which so that transpiration after sunset is virtually negligible.
 4 B. Bueno et al.
Roth (2007) shows this effect by comparing heat fluxes
measured in different cities with different vegetation
fractions. Other studies also indicate that the evapo-
transpiration from vegetation is mainly dependent on
solar radiation (Shashua-Bar and Hoffman 2002). The
difficulty is then to determine the fraction of absorbed
solar radiation that is converted into latent heat by
plants. As a first approximation, the current version of
the model assumes 50% conversion.
2.2. Vertical Diffusion Model
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012
The VDM reads air temperatures and velocities measured
at the weather station, as well as sensible heat fluxes
calculated by the RSM, and calculates vertical profiles of
air temperature above the weather station (Figure 3),
which are then provided to the UBL model.
The VDM solves the following heat diffusion
equation:
 
@yðzÞ 1 @ @yðzÞ
¼ rðzÞKd ðzÞ ; ð4Þ
@t rðzÞ @z @z
where y is the potential temperature of the air, which is
a variable commonly used in meteorology, represent-
ing the temperature that a parcel of fluid at a certain
pressure would acquire if is brought adiabatically to a
standard reference pressure. In Equation (4), z is the
vertical space component, r is the air density and Kd is
a diffusion coefficient. The lower boundary condition
of Equation (4) is the temperature measured at the
weather station, yrur(z ¼ 2 m). The upper boundary
condition accounts for the fact that at a certain height
(zref * 150 m), the profile of potential temperature is
Figure 3. Representation of a city and the physical domain
of the different UWG modules. The diagram corresponds to
nighttime conditions (not at scale). The RSM calculates
sensible heat fluxes based on a energy balance at the rural
soil. The VDM calculates vertical profiles of potential
temperature (yrur) at the rural site. The UBL model solves
an energy balance at the urban boundary layer between the
blending height (zr) and the boundary-layer height (zi) to
calculate air temperatures above the urban canopy layer. The
UC-BEM calculates air temperature and humidity inside
urban canyons.
uniform and @y
@z jzref ¼ 0. The diffusion coefficient is
calculated dynamically by the following equation
(Bougeault and Lacarrere 1989):
Kd ¼ Ck lk E1=2 ; ð5Þ
where E is the turbulent kinetic energy of the flow, Ck
is a model parameter set equal to 0.4 and lk is a length
scale. The turbulent kinetic energy at each vertical level
is approximated by:
 
E ¼ max w2s ; Emin ; ð6Þ
where ws is a mixed-layer velocity scale (Hong et al.
2006) and Emin is set equal to 0.01 m2 s72.
2.3. Urban Boundary-Layer model
The UBL model calculates air temperatures above the
urban canopy layer from the temperatures at different
heights provided by the VDM and the sensible heat
fluxes provided by the RSM and the UC-BEM.
The model is based on an energy balance for a
selected control volume inside the urban boundary
layer delimited by the blending height (zr) and the
boundary-layer height (zi) (Figure 3). The energy
balance is expressed as:
Z
dyurb
VCV rcv ¼ Hurb þ uref rcp ðyref  yurb ÞdAf ; ð7Þ
dt
where VCV is the control volume, r is the air density,
cv is the air specific heat at constant volume, cp is the
air specific heat at constant pressure, yurb is the
average potential temperature of the control volume,
Hurb is the sensible heat flux at the surface of the
control volume [W], yref is a reference potential
temperature outside the control volume, uref is a
reference air velocity and Af is the lateral area of heat
exchange between the control volume and its sur-
roundings. In Equation (7), the term on the left hand
side represents the thermal inertia of the control
volume and the second term on the right hand side
represents the advection effect. The model assumes
that the potential temperature is uniform inside the
control volume.
The UBL model differentiates between nighttime
and daytime urban boundary layers, and between the
advection effect driven by a geostrophic wind (forced
problem) and by the urban breeze circulation
(buoyancy-driven problem). The problem is assumed
to be buoyancy-driven if the circulation velocity is
greater than the air velocity measured at the weather
station. The circulation velocity (ucirc) is calculated
 Journal of Building Performance Simulation 5

from the expression proposed by Hidalgo et al. the sensible heat exchange between the canyon air and
(2009): the atmosphere, the heat fluxes due to exfiltration, the
  waste heat from heating, ventilation and air-condition-
Hurb  Hrur 1=3
ucirc ¼ kw bzi ; ð8Þ ing (HVAC) equipment and other anthropogenic heat
rcp sources, and the radiant heat exchange between the
canyon air and the sky. Thus, the urban canyon energy
where kw is a constant (kw * 1), b is the buoyancy balance is given by:
coefficient (b ¼ gy71) and Hurb and Hrur are the
sensible heat fluxes [W m72] from the urban and the dTurb
Vcan rcv ¼ Aw hw ðTw  Turb Þ
rural sites, respectively. dt ð10Þ
The numerical method used to solve Equation (7) is þ Ar hr ðTr  Turb Þ þ Ar hrd;sky ðTsky  Turb Þ
y y
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012

implicit Euler, in which dydturb ¼ urb d urb , where d is the

simulation timestep. Then, Equation (7) can be Awin Uwin ðTin  Turb Þ þ V_ inf=vent rcp ðTin  Turb Þ
expressed as: þ uex rcp ðTubl  Turb Þ þ Hwaste þ Htraffic ;
yurb  y
urb ¼ Csurf þ Cadv yeq  Cadv yurb ; ð9Þ
where Csurf, Cadv and yeq are calculated for each where Turb, Tin and Tubl are the air temperature of the
scenario according to Table 1. urban canyon, the indoor air temperature and the air
temperature of the urban boundary layer above the
urban canyon, respectively; Tsky is the effective sky
2.4. Urban Canopy and Building Energy Model temperature; Vcan is the volume of the urban canyon
The UC-BEM calculates urban canyon air temperature air; Uwin is the U-factor of windows including heat
and humidity from radiation and precipitation data, exchange coefficients at both sides; V_ inf=vent is the
air velocity and humidity measured at the weather exfiltration airflow rate; Hwaste is the sensible compo-
station, and from the air temperature above the urban nent of waste heat flux released by HVAC systems into
canopy layer calculated by the UBL model. the urban canyon; Htraffic represents other anthropo-
The model is based on the town energy balance genic sources of heat and uex is the exchange velocity
(TEB) scheme (Masson 2000), including its building between the in-canopy and above-canopy flows. An
energy model (Bueno et al. 2012b). The UC-BEM analogous latent heat balance is solved to calculate the
assumes that the air inside the urban canopy layer is humidity content of the urban canyon air by comput-
well mixed. Urban canyon air temperatures are ing the latent heat fluxes from the atmosphere,
obtained by the heat balance method, taking into buildings and road. The UWG assumes that the air
account the heat capacity of the urban canyon air humidity above urban canyons is the same as the one
(Figure 1). The urban canyon energy balance accounts measured at the weather station for each time step.
for the heat fluxes from walls, windows and the road, The exchange velocity (uex) is obtained from an
expression extracted from Bentham and Britter (2003):
Table 1. Surface coefficient (Csurf), advection coefficient u
uex ¼  1=2 ; ð11Þ
(Cadv) and equivalent temperature (yeq) used in Equation (9) uatm 8
for each scenario. For nighttime-forced scenario, the urban u  VHurb
boundary layer is horizontally divided in various control
volumes. yrur is the potential temperature outside the city at
different heights {zr, zi and zref}. yn71 is the average potential where u* is the friction velocity (see Appendix 1) and
temperature of the control volume upstream of the one
considered. D, Acity and Pcity are the city diameter, horizontal
uatm is a reference air velocity above the urban canopy
area and perimeter, respectively; and uwind is the air velocity assumed equal to the air velocity measured at the
measured at the weather station. weather station.
In large spaces such as urban canyons, the
Csurf Cadv yeq
water vapour present in the air participates in the
Night radiant heat exchange. The air emissivity is calculated
2
Forced (first) Hurb d
zi rcv
uwind ðzm Þzi dcp
3 yrur ðzi Þ þ 13 yrur ðzr Þ as a function of the humidity content and the size of
2zm dxcv
Forced (rest) yn71 the space (Siegel and Howell 1981). In Equation (10),
Buoyancy-driven Hurb d Pcity ucirc dcp 1 1
zi rcv Acity cv 2 yrur ðzi Þ þ 2 yrur ðzr Þ hw and hr are the heat transfer coefficients of walls and
Day road, respectively, which combine convective and
Forced Hurb d Wuwind ðzm Þdcp yrur(zref)
zi rcv Acity cv
radiative effects (h ¼ hcv þ hrd); and hrd,sky is the
Buoyancy-driven Hurb d Pcity ucirc dcp yrur(zref) radiant heat transfer coefficient between the urban
zi rcv Acity cv
canyon air and the sky (see Appendix 2).
 6 B. Bueno et al.

External surface temperatures of walls, road and
roof are calculated by solving a similar surface energy
balance to the one described for the rural soil
(Equations (1)–(3)). The boundary conditions of the
road are the same as for the rural soil. In the case of
walls and roof, the indoor boundary condition is a heat
flux calculated by the building energy model.
The outdoor surface heat flux is composed of
shortwave radiation, longwave radiation, sensible and
latent heat components. The solar radiation received
by walls and road is calculated by assuming an average
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012
For horizontal surfaces (road and roof), the UC-
BEM calculates the latent heat flux associated with a
thin layer of water (max 1 mm) which remains after the
precipitation water is run off. The mass balance to the
film of water is given by:
dwg
¼ ðPg  Eg Þ; ð14Þ
dt
where Pg and Eg are the precipitation and evaporation
mass fluxes, respectively [m s71]. The water evapora-

urban canyon orientation (see Appendix 3). The
longwave radiation among walls, road, urban canyon
air and the sky is computed by linearization of the
Stefan–Boltzmann equation accounting for the trans-
mittance of the urban canyon air and assuming only
one bounce of radiative heat fluxes between surfaces
(see Appendix 2). In terms of longwave radiation,
window surfaces are assumed to have the same
temperature as wall surfaces. Surface sensible heat
fluxes are computed by using CHTC, which are
calculated as a function of the air velocity above the
urban canopy layer (uatm) by using a correlation
extracted from Palyvos (2008):
hcv ¼ 5:8 þ 3:7uatm :
The air velocity inside the urban canyon (ucan) is
used in Equation (12) for the road. This is given by the
following equation (Bentham and Britter 2003):
 1=2
8 balance of the urban canyon. Although the tree canopy
ucan ¼ u : ð13Þ
VHurb
Urban sensible heat fluxes (required by the UBL
model) are calculated as the sum of the heat exchange
between the canyon air and the atmosphere and the
convective heat flux from building roofs, including the
fraction of waste heat emissions from outdoor HVAC
equipment located there.
tion is calculated as:
1
Eg ¼ ðqsat ðTsurf Þ  qa Þ; ð15Þ
rw R
where R is an aerodynamic resistance [s m71] obtained
from the CHTC (Equation (12)), rw is the water
density, qsat is the saturation specific humidity [kg
kg71] at surface temperature and qa is the specific
humidity of the air above the surface. Whenever the
depth of the layer of water is greater than zero, the
latent heat flux is calculated as L ¼ Eg rwlv, where lv is
the latent heat of vapourization [J kg71].
The vegetation model of the UC-BEM follows the
shade-convection approach (Shashua-Bar and Hoff-
ð12Þ man 2002). The solar radiation that reaches urban
canyons is partially blocked by the tree canopy
according to the horizontal vegetation density of the
site. The solar radiation absorbed by the trees is split
into sensible and latent heat fluxes as indicated in
Section 2.1. These fluxes then participate in the energy
reduces the sky view factor of the urban canyon and
consequently may reduce the net outgoing radiation,
the model assumes that the temperature of urban
surfaces is close enough to the one of the tree canopy
so that the overall effect of trees on the longwave
radiation balance is negligible.
The building energy model is based on the one
developed by Bueno et al. (2012b). The physical and

Table 2. Monthly normals of climate variables in Basel, Switzerland and Toulouse, France.

Variable Site Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Min temperature (8C) Basel 72 71 2 4 8 11 13 13 10 7 2 71
Toulouse 2 3 4 7 10 13 15 15 13 10 5 3
Max temperature (8C) Basel 4 6 10 14 18 22 24 23 20 15 8 4
Toulouse 9 11 13 16 20 24 27 27 24 19 13 9
Mean temperature (8C) Basel 1 2 6 9 13 16 19 18 15 10 5 2
Toulouse 5 7 9 11 15 18 21 21 19 14 9 6
Mean wind speed (m s–1) Basel 3 3 3 2 2 2 2 2 2 3 3 3
Toulouse 3 3 3 3 3 3 3 2 2 2 2 3
Mean precipitation (mm) Basel 53 40 48 55 76 93 86 94 79 60 56 44
Toulouse 55 55 58 64 73 58 41 47 48 52 49 56

Sources: MeteoSwiss (www.meteoschweiz.ch), Info Climat (www.infoclimat.fr) and Climate Temp (www.climatetemp.info).
 Journal of Building Performance Simulation 7

Table 3. System parameters of the UWG used in the model Table 4. Inputs of the UWG used in the model comparison
comparison with field data from Basel, Switzerland and with field data from Basel, Switzerland and Toulouse,
Toulouse, France. France. Both parameterizations represent densely populated
residential areas and rural areas covered by grass. The
Parameter Setting thermal properties of construction materials are indicated in
Table 5.
Daytime boundary-layer height 1000 m
Nighttime boundary-layer height 50 m Parameter BUBBLE CAPITOUL
Reference height at which the vertical 150 m
profile of potential temperature Urban parameters
is assumed uniform Location Basel Toulouse
Urban-breeze scaling coefficient 1.2 Latitude 47.338 43.488
Latent fraction of vegetation 0.5 Longitude 7.358 1.38
Albedo of vegetation 0.25 City diameter 5000 m 7500 m
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012

Begin month for vegetation participation May Average building 14.6 m 20 m

End month for vegetation participation November height
Horizontal 0.54 0.68
building density
Vertical-to-horizontal 0.48 1.10
geometric definition of buildings is kept as simple as urban area ratio
possible, while maintaining the required features of a Horizontal vegetation 0.16 0.08
density (trees)
comprehensive building energy model. The model Wall construction Concrete –20 cm Brick – 30 cm
considers a single thermal zone, where the thermal Insulation – 3 cm Insulation
inertia of building materials associated with multiple – 3 cm
levels is represented by a generic thermal mass. The Wall albedo 0.15 0.25
Roof construction Tiles – 6 cm Tiles – 6 cm
model accounts for heat gains due to transmitted solar Concrete – 20 cm Wood – 20 cm
radiation, heat conduction through the enclosure, Insulation – 3 cm Insulation
infiltration, ventilation and internal heat gains, as well – 3 cm
as for the dynamical evolution of indoor air tempera- Roof albedo 0.15 0.25
Road construction Asphalt – 5 cm Asphalt – 5 cm
ture (between thermal set points) and humidity. Stones – 20 cm Stones – 20 cm
To calculate cooling energy consumption, the Gravel and Gravel and
model solves the dehumidification of the air passing soil – 100 cm soil – 100 cm
Road albedo 0.08 0.08
through the cooling system by assuming that the air
leaves the cooling coil at 90% relative humidity. The Building parameters
Building floor Concrete Concrete
model includes the mixture of recirculated air and construction – 20 cm – 20 cm
outdoor air according to the ventilation air flowrate. Glazing ratio 0.3 0.3
Waste heat fluxes are calculated as a function of the Window Double-pane Double-pane
construction clear glass clear glass
building energy consumption (Qcons) and building Internal heat gains Residential Residential
energy demand (Qdem). For example, for a cooling (5.8 W m–2)
system the waste heat flux is given by: Infiltration/ 0.5 ACH 0.5 ACH
ventilation
Qwaste ¼ Qcons þ Qdem : ð16Þ Cooling system None None
Heating system Furnace Furnace
Weather station
parameters
3. Model evaluation Construction Soil Soil
The UWG scheme is compared with field data from Albedo of the 0.15 0.15
surface without
two boundary-layer experiments: the intensive obser- vegetation
vational period (IOP) of the BUBBLE experimental Vegetation fraction 0.8 0.8
campaign, carried out in Basel (Switzerland) between
10 June and 10 July 2002 (Rotach et al. 2005); and the
CAPITOUL experimental campaign carried out in The air temperature measurement heights for the
Toulouse (France) from February 2004 to March 2005 case of CAPITOUL are 19 m and 2 m for the urban
(Masson et al. 2008). The climate information of both and the rural sites, respectively. The measurement
sites is characterized in Table 2. heights for the case of BUBBLE are 2.6 m and 1.5 m.
In both experiments, weather data are measured Air velocity was measured at 47.5 m above the urban
simultaneously at rural and urban stations. The site for the case of CAPITOUL and at 28 m above the
evaluation of the UWG consists of introducing rural rural site for the case of BUBBLE. The model assumes
air temperatures as inputs in the model and comparing that the wind speed measured above the surface layer
the calculated and observed urban air temperatures. at a rural site (above 10 m) is similar to the wind speed
 8 B. Bueno et al.

Table 5. Thermal properties of construction materials used Table 6. Root-mean-square error (RMSE) and mean-bias
in the simulations. error (MBE) between the urban air temperatures calculated
by the UWG and observed during BUBBLE experiment
Thermal Volumetric between 10 June and 10 July, 2002; and between the urban
conductivity heat capacity air temperatures calculated by the UWG and observed
Material (W m–1 K–1) (J m–3 K–1) during CAPITOUL experiment in July and October, 2004,
and January, 2005. Errors are compared to the average daily-
Brick and tiles 1.15 1.6e6 maximum UHI effect (UHImax ) observed during each period.
Concrete 0.93 1.5e6
Wood 0.93 1.2e6 Month RMSE (K) MBE (K) UHImax (K)
Insulation 0.03 5.2e4
Asphalt 0.74 1.9e6 BUBBLE
Stones 2.10 2.0e6 Summer 0.9 70.6 5.2
Gravel and soil 0.40 1.4e6 CAPITOUL
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012

Summer 0.7 0.2 3.6

Fall 0.8 70.1 2.5
Winter 1.1 70.2 2.4

parameters of the VDM-UBL scheme are the daytime

Figure 4. Hourly values of urban canyon air temperature
calculated by the UWG and observed during the BUBBLE
experiment between 21 June and 28 June 2002. Hourly values
of measured rural air temperature (rur) for the same period
are also represented.
Figure 5. Monthly-average diurnal cycle of urban canyon
air temperature calculated by the UWG and observed during
the BUBBLE experiment between 10 June and 10 July 2002.
Monthly-average diurnal cycle of measured rural air
temperature (rur) for the same period is also represented.
above the urban canopy layer. This can be understood
by following a stream line from a rural site that is
deformed when passes over a city (Figure 3).
The system parameters of the UWG used in this
comparison are summarized in Table 3. The
and nighttime boundary layer heights (zi), the reference
height (zref) and the urban-breeze scaling coefficient
(kw) (Bueno et al. 2012a). In addition, the sensible-
latent heat split of vegetation is assumed to be 0.5 and
the vegetation albedo is taken as 0.25, which is the
average value reported in the experiments. Finally, the
model accounts for the effect of vegetation from May
to November (deciduous vegetation).
3.1. Comparison with field data from Basel,
Switzerland
The main urban experimental site in BUBBLE is Basel-
Sperrstrasse. The site represents a heavily built-up part
of the city centre of Basel, mainly composed of
residential buildings. Tables 4 and 5 show the inputs
parameters used in the evaluation of the UWG, based
on a characterization of the urban site carried out by
Hamdi and Masson (2008). A sensitivity analysis of
input parameters is presented in Section 3.3.
The Grenzach weather station, inside the valley of
the Rhine River, is used as the reference rural station.
Figure 4 compares hourly values of urban air
temperatures calculated by the UWG with the air
temperatures measured at the urban and rural sites for
a week in June. The monthly-average diurnal cycle for
the IOP of the BUBBLE campaign is represented in
Figure 5. As can be seen, the UWG is able to capture
both the UHI effect observed at night and the UCI
effect observed during the day. The underprediction of
urban air temperature can be partially explained by the
fact that the measurement height is 2.6 m, which might
capture a warmer air than the average canopy
temperature calculated by the model.
Statistical results of this comparison are presented
in Table 6. The root-mean-square error (RMSE)
 Journal of Building Performance Simulation 9
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012

Figure 6. Map of the weather station network during the CAPITOUL experiment carried out in Toulouse, France, from
February 2004 to March 2005.

between the model and observations is 0.9 K, where

the average daily-maximum UHI effect is 5.2 K. Errors
of about 1 K are acceptable given the important
uncertainties associated with urban climate predictions
(see next section). The mean-bias error (MBE) is
70.6 K, which reproduces the underprediction of
urban air temperature observed in Figure 5.

3.2. Comparison with field data from Toulouse, France

The CAPITOUL campaign is an extensive boundary-
layer experiment, which includes (among other types of
measurements) a network of weather stations inside
and at the periphery of Toulouse (Figure 6). In this
analysis, the station located at the central location of
the city, next to the Monoprix building (MNP), is
selected as representative of urban conditions. Five of
the surrounding urban stations are also included in the
analysis (MIC, CIT, MIN, ILE and CYP) to show the
variability of air temperatures within the same urban
area. Table 4 summarizes the inputs parameters of the
model, based on a previous characterization of the
urban site (Bueno et al. 2011). The reference weather
station is located at Mondouzil (MON), an agricultur-
al rural area at the North-East periphery of the city.
Figure 7 compares hourly values of urban air
temperatures calculated by the UWG with the air
temperatures measured at the urban and rural sites for
a week in July. The monthly-average diurnal cycles for
Figure 7. Hourly values of urban canyon air temperature
calculated by the UWG and observed during the
CAPITOUL experiment between 21 July and 28 July 2004.
Hourly values of measured rural air temperature (rur) for the
same period are also represented.
July and October, 2004, and January, 2005, are
represented in Figure 8. The overprediction of urban
air temperature in the morning can be related to the
measurement height (19 m), which may capture a
cooler air than the average over the urban canopy.
The error bar represents the root-mean-square
difference between the air temperatures observed in
the five urban stations surrounding the MNP station
and the air temperature measured at the MNP station
(Figure 6). The results show the capacity of the UWG
 10 B. Bueno et al.

Table 7. Maximum variation of the RMSE or the MBE

obtained in the comparison between the urban air tempera-
tures calculated by the UWG and observed during the
experiments in summer when model parameters range
between 725% and þ25% of the reference value indicated
in Tables 3 and 4. Units are K.

Parameter BUBBLE CAPITOUL

City diameter 0.1 0.1
Average building height 0.1 0.1
Horizontal building density 0.4 0.4
Vertical-to-horizontal 0.8 0.3
urban area ratio
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012

Horizontal vegetation 0.1 0.1

density (trees)
Wall albedo 0.1 0.1
Roof albedo 0.1 0.1
Road albedo 0.1 0.1
Volumetric heat capacity 0.1 0.1
of concrete/brick in walls
Volumetric heat capacity 0.2 0.1
of asphalt in road
Internal heat gains 0.1 0.1
Rural vegetation fraction 0.3 0.1
Daytime mixing height 0.1 0.1
Nighttime boundary- 0.1 0.1
layer height
Reference height at which the 0.1 0.1
vertical profile of potential
temperature is assumed
uniform
Urban-breeze scaling coefficient 0.1 0.1
Latent fraction of vegetation 0.4 0.1

1.1 K, where the average daily-maximum UHI effect

Figure 8. Monthly-average diurnal cycles of urban canyon
air temperature calculated by the UWG and observed during
the CAPITOUL experiment in July (top) and October
(middle), 2004, and in January (bottom), 2005. Monthly-
average diurnal cycles of measured rural air temperature
(rur) for the same periods are also represented. The error bar
represents the root-mean-square difference between the air
temperatures observed in the five urban stations surrounding
the MNP station and the air temperature measured at the
MNP station.
to reproduce the UHI effect for different seasons. The
observed variability of air temperature around the
MNP weather station is about 1 K. This justifies the
statement that the error associated with UWG’s
predictions is acceptable and within the air tempera-
ture range observed in different locations of the same
urban area.
Statistical results of this comparison are presented
in Table 6 for the three months. The RMSE between
the model and observations ranges between 0.7 K and
ranges between 2.4 K and 3.6 K. The MBE is generally
low, which indicates that there are no systematic errors
in the model.
Observations show that the UHI effect at mesoscale
level (due to the aggregate effects of the whole city)
cannot be neglected. From the daily-maximum UHI
effect observed inside urban canyons, more than half is
due to the mesoscale effect.
3.3. Sensitivity analysis
A sensitivity analysis is carried out for some of the
model parameters. Table 7 shows the variation of the
error obtained in the comparison between the UWG
and observations in summer when model parameters
range between 725% and þ25% of the reference
value indicated in Tables 3 and 4. It can be seen that,
although all parameters have an effect, the most critical
ones are the morphological parameters; in particular,
the horizontal building density and the vertical-to-
horizontal urban area ratio. Values for these para-
meters are available for numerous cities around the
world and can be obtained from geographic informa-
tion systems.
 Journal of Building Performance Simulation 11

The case study of BUBBLE is sensitive to some vege- Institute of Science. Researchers from the GAME/CNRM
tation parameters for which it is difficult to find exact (Météo-France, CNRS) were supported by the French
National Research Agency (ANR) under grant ANR-09-
values. This indicates an area for model improvement. VILL-0003 and the Scientific Cooperation Foundation STAE
in Toulouse in the context of the MUSCADE and the
ACCLIMAT projects, respectively. We would also like to
4. Conclusion thank Dr. Roland Vogt for sharing the BUBBLE data with us.
This paper presents a computationally fast model, based
on energy conservation principles, to predict the UHI
effect in a city given meteorological information mea- Nomenclature
sured in an operational weather station outside the city. Symbol Designation Unit
The UWG has been satisfactorily evaluated against Acity City horizontal area m2
m2
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012

field data from two cities: Basel, Switzerland and Af Lateral heat exchange area
cp Air specific heat at J kg71 K71
Toulouse, France. The expected error associated with constant pressure
UWG predictions is about 1 K, which stays within the cv Air specific heat at J kg71 K71
range of air temperature variability observed in different constant volume
locations of the same urban area. The comparison with C Coefficients/thermal /W m72 K71
conductance
field data highlights that the UHI effect cannot be Cvk Von-Karman constant –
computed only from the urban canyon effect (vertical d Layer thickness m
component), but must also include the aggregate effect D City diameter m
of the whole city (horizontal component). As a E Turbulent kinetic energy m2 s72
of the flow
consequence, urban climate prediction tools cannot be Eg Evaporation mass flux m s71
limited to an urban canopy model, but must also F View factor –
consider the effect of the urban boundary layer. This can g Gravity acceleration m s72
be achieved by using mesoscale atmospheric simulations h Heat transfer coefficient W m72 K71
hbld Average building height M
or by using the simplified approach of the VDM-UBL H Sensible heat flux W m72
scheme (Bueno et al. 2012a). kw Urban-breeze circulation –
The reference weather station for the UWG can be scale constant
situated in any location on the periphery of the city as K Solar radiation W m72
Kd Diffusion coefficient m2 s71
long as it is not surrounded by urbanization and is not lk Length scale M
affected by site-specific micro-climate conditions pro- lv Water condensation heat J kg71
duced by the orography or by the presence of large L Latent heat flux W m72
P Partial pressure Pa
bodies of water. For example, a weather station near Pcan Canyon perimeter m
the sea would not be appropriate for applying the Pcity City perimeter m
UWG. The current version of the UWG has performed Pg Precipitation mass flux m s71
well in European-type cities in which the urban q Specific humidity kg kg71
Q Heat flux W m72
morphology is relatively homogeneous and the urban R Aerodynamic resistance m s71
vegetation is scarce. Further developments of the Ri Richardson number –
model will address the heterogeneity of urban areas S Received solar radiation W m72
and the spatial distribution of the UHI effect within a t Time S
T Temperature 8C, K
city. They will also include a better treatment of latent u Mean air velocity m s71
heat fluxes, while maintaining the approach of keeping uex Exchange velocity m s71
the model as simple as possible. u* Friction velocity m s71
The UWG can be integrated into existing pro- Uwin Window U-factor W m72 K71
V Volume m3
grammes in order to account for site-specific urban V_ Air volume flowrate m3 s71
weather files in building energy simulations. It can also VHurb Vertical-to-horizontal –
be converted into a fully-operative programme to urban area ratio, defined
predict building energy consumption at urban scale as exterior vertical building
area divided by the plan
and the UHI effect, taking into account the energy area of the urban site
interactions between buildings and the urban climate. wg Water thickness m
wr Average road width m
ws Mixed-layer velocity scale m s71
Acknowledgements z Vertical space component m
This research was funded by the Singapore National Research zi Boundary-layer height m
Foundation through the Singapore-MIT Alliance for Re- zr Blending height m
search and Technology (SMART) Centre for Environmental zref Reference height m
Sensing and Modelling (CENSAM) and the MIT/Masdar z0 Roughness length m
 12 B. Bueno et al.

b Buoyancy coefficient m s71 K71 Erell, E. and Williamson, T., 2006. Simulating air temperature in
d Timestep s an urban street canyon in all weather conditions using
e Emissivity – measured data at a reference meteorological station.
y Potential temperature K International Journal of Climatology, 26 (12), 1671–1694.
yo Critical canyon orientation – Erell, E. and Williamson, T., 2007. Intra-urban differences in
s Stefan-Boltzmann constant W m72 K74 canopy layer air temperature at a mid-latitude city.
l Solar zenith angle – International Journal of Climatology, 27 (9), 1243–1255.
r Density kg m73 Grimmond, C.S.B. and Oke, T.R., 2002. Turbulent heat fluxes in
rbld Building density, defined as –
urban areas: observations and a local-scale urban meteor-
building plan area divided
ological parameterization scheme (LUMPS). Journal of
by the plan area of
the urban site Applied Meteorology, 41 (7), 792–810.
rr Road reflectivity – Hamdi, R. and Masson, V., 2008. Inclusion of a drag approach
rw Wall reflectivity/water density –/kg m73 in the town energy balance (TEB) scheme: offline 1D
evaluation in a street canyon. Journal of Applied Meteorol-
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012

Subscript
ogy and Climatology, 47 (10), 2627–2644.
a Air
atm Atmosphere Hicks, B.B., Callahan, W.J., and Hoekzema, M.A., 2010. On the
bld Building heat islands of Washington, DC, and New York City, NY.
can Urban canyon Boundary-Layer Meteorology, 135 (2), 291–300.
cons HVAC consumption Hidalgo, J., Masson, V., and Gimeno, L., 2009. Scaling the
cv Convection daytime urban heat island and urban-breeze circulation.
dem Building demand Journal of Applied Meteorology and Climatology, 49 (5), 889–
dif Diffuse 901.
dir Direct Hong, S.Y., Noh, Y., and Dudhia, J., 2006. A new vertical
in Indoor air diffusion package with an explicit treatment of entrainment
inf Infiltration processes. Monthly Weather Review, 134 (9), 2318–2341.
r Road Houet, T. and Pigeon, G., 2011. Mapping urban climate
rd Radiation zones and quantifying climate behaviors – an application
ref Reference on Toulouse urban area (France). Environmental Pollu-
rur Rural tion, 159, 2180–2192.
sat Saturation Kikegawa, Y., et al., 2003. Development of a numerical
ubl Urban boundary layer simulation system toward comprehensive assessments of
urb Urban urban warming countermeasures including their impacts
vent Ventilation upon the urban buildings’ energy-demands. Applied
w Walls
Energy, 76 (4), 449–466.
waste Waste heat from
HVAC systems Kuttler, W., 2008. The urban climate – basic and applied
win Windows aspects. Urban Ecology, 233–248.
Lee, S.H. and Baik, J.J., 2010. Statistical and dynamical
characteristics of the urban heat island intensity in
Seoul. Theoretical and Applied Climatology, 100, 227–237.
References Louis, J.F., 1979. A parametric model of vertical eddy fluxes in
Bentham, T. and Britter, R., 2003. Spatially averaged flow the atmosphere. Boundary-Layer Meteorology, 17, 187–202.
within obstacle arrays. Atmospheric Environment, 37 (15), Masson, V., 2000. A physically-based scheme for the urban
2037–2043. energy budget in atmospheric models. Boundary-Layer
Bougeault, P. and Lacarrere, P., 1989. Parameterization Meteorology, 94, 357–397.
of orography-induced turbulence in a mesobeta-scale Masson, V., et al., 2008. The canopy and aerosol particles
model. Monthly Weather Review, 117, 1872–1890. interactions in TOulouse Urban Layer (CAPITOUL)
Bueno, B., et al., 2011. Combining a detailed building energy experiment. Meteorology and Atmospheric Physics, 102,
model with a physically-based urban canopy model. 135–157.
Boundary-Layer Meteorology, 140, 471–489. Noilhan, J. and Mahfouf, J.F., 1996. The ISBA land surface
Bueno, B., et al., 2012a. Air temperatures above the parameterisation scheme. Global and Planetary Change,
urban canopy layer from measurements at a rural 13, 145–159.
operational weather station. Journal of Applied Meteorol- Oke, T., 1987. Boundary layer climates. London: University
ogy and Climatology (accepted). Paperbacks Methuen.
Bueno, B., et al., 2012b. Development and evaluation of a Oxizidis, S., Dudek, A., and Papadopoulos, A., 2008. A
building energy model integrated in the TEB scheme. computational method to assess the impact of urban
Geoscientific Model Development, 5 (2), 433–448. climate on buildings using modeled climatic data. Energy
Bueno, B., et al., 2012c. A resistance-capacitance network and Buildings, 40 (3), 215–223.
model for the analysis of the interactions between the Palyvos, J., 2008. A survey of wind convection coefficient
energy performance of buildings and the urban climate. correlations for building envelope energy systems’ model-
Building and Environment, 54, 116–125. ing. Applied Thermal Engineering, 28 (8–9), 801–808.
Crawley, D.B., 2008. Estimating the impacts of climate change Rotach, M.W., et al., 2005. BUBBLE an urban boundary
and urbanization on building performance. Journal of layer meteorology project. Theoretical and Applied
Building Performance Simulation, 1 (2), 91–115. Climatology, 81, 231–261.
Crawley, D.B., et al., 2001. EnergyPlus: creating a new- Roth, M., 2007. Review of urban climate research in
generation building energy simulation program. Energy (sub)tropical regions. International Journal of Climatol-
and Buildings, 33 (4), 319–331. ogy, 27 (14), 1859–1873.
 Journal of Building Performance Simulation 13

Sailor, D.J., 2011. A review of methods for estimating The air emissivity (ecan) is calculated from the following
anthropogenic heat and moisture emissions in the urban expression as a function of the humidity content and the size
environment. International Journal of Climatology, 31 (2), of the space (Siegel and Howell 1981):
189–199.   
Shashua-Bar, L. and Hoffman, M.E., 2002. The Green ecan ¼ 0:683 1  exp 1:17X1=2 ; ðA7Þ
CTTC model for predicting the air temperature in small
urban wooded sites. Building and Environment, 37 (12), where X ¼ Pw Le T300 ðPa þ bPw Þ; and where Pw and Pa are the
1279–1288. urb

Siegel, R. and Howell, J., 1981. Thermal radiation heat partial pressures of water and air, respectively; Le is the mean
transfer. Series in thermal and fluids engineering. beam length calculated as the ratio between the transversal
London: Hemisphere Publishing Corporation. area of the urban canyon and  its  perimeter,
1=2
300
Swaid, H. and Hoffman, M.E., 1990. Prediction of urban air Le ¼ 3:6Acan =Pcan and the parameter b ¼ 5 Turb :
temperature variations using the analytical CTTC model.
Energy and Buildings, 14 (4), 313–324.
Downloaded by [Massachusetts Institute of Technology], [Bruno Bueno] at 12:26 31 August 2012

Zhou, Y. and Shepherd, J., 2010. Atlanta’s urban heat island

under extreme heat conditions and potential mitigation Appendix 3. Solar radiation received by walls and roofs
strategies. Natural Hazards, 52, 639–668. This formulation has been extracted from Masson (2000) for
an average oriented urban canyon. The first bounce of solar
radiation is given by the following equations for walls and
roof:
Appendix 1. Friction velocity
The friction velocity, u*, is calculated according to Louis    
(1979): wr yo 1
Kw ¼ Kdir 0:5  þ tanðlÞð1  cosðyo ÞÞ þ Fw Kdif ;
hbld p p
1=2
u ¼ auatm fm ; ðA1Þ ðA8Þ
where a ¼ Cvk
z is a drag coefficient, Cvk ¼ 0.4 is the Von-  
logz r 2yo 2 hbld
0
Karman constant, zr is the blending height taken as Kr ¼ Kdir þ tanðlÞð1  cosðyo ÞÞ þ Fr Kdif ; ðA9Þ
p p wr
zr ¼ 1.5hbld (at the urban site), z0 is the roughness length
taken as z0 ¼ hbld/10 and fm is a coefficient that accounts for
atmosphere stability and is given by: where Kdir is the direct solar radiation; Kdif is the diffuse solar
radiation; hbld is the average building height; wr is the average
ð1rbld Þ
fm ¼
1
; ðA2Þ road width, wr ¼ 2hbldVH urb
; rbld is the horizontal building
ð1 þ 4:7RiÞ2 density; VHurb is the vertical-to-horizontal urban area
ratio; l is the solar zenith angle; yo is the critical canyon
for stable and neutral conditions (Ri  0) and orientation forh which the i road in no longer sunlit,
wr 1
yo ¼ arcsin min hbld tanðlÞ ; 1 ; Fw is the wall sky view
1  9:4Ri   1=2
fm ¼ ; ðA3Þ 1 hbld hbld 2
1 þ cðRiÞ2 factor, Fw ¼ 2 wr þ 1  ð wr Þ þ 1 = hwbldr and Fr is the
 2
1=2
for unstable conditions Ri 5 0. In Equation (A3), road sky view factor, Fr ¼ ðhwbldr Þ þ 1  hwbldr .
 1=2
the constant c is given by c ¼ 69:56a2 zz0r ; and the Solar reflections are then calculated by:
Richardson number is calculated as:
Rw þ Fw rw Rr
gzr ðyubl  ycan Þ Mw ¼ ; ðA10Þ
Ri ¼ : ðA4Þ 1  ð1  2Fw Þrw þ ð1  Fr ÞFw rr rw
yubl u2atm

and
Appendix 2. Outdoor radiant heat transfer coefficients
The radiant heat transfer coefficients (RHTC) among walls, Rr þ ð1  Fr Þrr ðRw þ Fw rw Rr Þ
Mr ¼ ; ðA11Þ
road and sky are given by: 1  ð1  2Fw Þrw þ ð1  Fr ÞFw rr rw
3
hrd;ij ¼ 4ð1  ecan Þei ej sFij Tij ; ðA5Þ
where rw and rr are the wall and road surface reflectivities,
where ecan is the urban canyon air emissivity (Equation Rw ¼ rwKw and Rr ¼ rrKr.
(A7)), e is the emissivity of the surfaces i,j ¼ {wall, road, sky Then, the total solar radiation received by walls and road
(esky ¼ 1)}; s ¼ 5.67e78Wm72K74 is the Stefan–Boltzmann is given by:
constant; Fi7j is the view factor between surfaces i and j and
Sw ¼ Kw þ ð1  2Fw ÞMw þ Fw Mr ; ðA12Þ
Tij is average temperature over surfaces i and j.
The RHTC between walls, road, sky and the urban
and
canyon air are given by:
3 Sr ¼ Kr þ ð1  Fr ÞMw : ðA13Þ
hrd;canj ¼ 4ecan ej sTcanj : ðA6Þ