Energy and Buildings 82 (2014) 100–113

Contents lists available at ScienceDirect

Energy and Buildings

journal homepage: www.elsevier.com/locate/enbuild

On the energy impact of urban heat island and global warming on

buildings
M. Santamouris a,b,∗
a
Physics Department, Univ. Athens, Greece
b
The Cyprus Institute, Nicosia, Cyprus

a r t i c l e i n f o a b s t r a c t

Article history: Urban heat island and global warming increase ambient temperature and modify the energy budget of
Received 3 July 2014 buildings. The magnitude of the modification has been evaluated in a large number of articles, under
Accepted 9 July 2014 different climatic and building boundary conditions. This paper collects, analyzes and classifies existing
Available online 18 July 2014
knowledge regarding the energy impact of urban heating to buildings and calculates preliminary indi-
cators and impact figures. Based on the analysis of the impact studies, it is found that in average the
Keywords:
cooling load of typical urban buildings is by 13% higher compared to similar buildings in rural areas. Four
Heat island
specific energy impact indicators, the global energy penalty per m2 , the global energy penalty per m2 and
Global warming
Energy consumption buildings
degree of UHI, the global energy penalty per person and the global energy penalty per person and per
Cooling energy degree of the UHI are defined and calculated. The variability of the heating and cooling loads of typical
buildings is evaluated for the period 1970–2010. The average increase of the cooling demand is 23% while
the corresponding average reduction of the heating is 19%. In total, the average energy consumption of
typical buildings for heating and cooling purposes increased by 11% for the same period.
© 2014 Elsevier B.V. All rights reserved.

1. Introduction of urban pollutants and in particular of the tropospheric ozone

Urban heat island and global warming increase the near surface
ambient temperature in cities [1]. Urban heat island is extremely
well documented and relevant studies exist for most of the major
cities in the world [2]. On the contrary few studies are available
on the specific impact of the global warming on the urban cli-
mate [3]. The magnitude of the urban heat island expressed as the
urban heat island intensity depends on many parameters like the
topographic characteristics of the city, the weather conditions, the
urban density, the anthropogenic heat released, the thermal and
optical characteristics of the used materials in the city fabric, the
urban form and land use while its is very influenced by the specific
characteristics of the selected reference – rural station [4].
Urban warming has serious consequences on the energy, envi-
ronmental and social balance of cities. The most serious impact
is associated to the significant increase of the peak and total
electricity demand especially for cooling purposes [5,6]. In par-
allel, urban warming is associated to increase the concentration
∗ Correspondence to: Physics Department, Univ. Athens, Greece.
Tel.: +30 2107276847; fax: +30 2107295282.
E-mail addresses: msantam@phys.uoa.gr, msantam@me.com
[7], worsen thermal comfort conditions in outdoor spaces [8,9],
enhance health and mortality problems [10,11], and significantly
increase the urban ecological footprint [12].
The magnitude of the specific energy impact of urban warming
depends on many parameters such as the intensity of the urban
overheating, the type and characteristics of the buildings under
consideration and the local microclimate. Increase of the urban
temperatures may exacerbate the energy demand for cooling pur-
poses; however, the heating needs may be reduced as well. The
specific energy impact of urban warming, under various climatic
boundary conditions and for many types of buildings, has been
analyzed in a wide number of studies. The methodology used, the
period of the analysis, as well as the outcome of the studies vary
substantially between the various studies, a fact which results in
difficulties regarding the intercomparison of results and the extrac-
tion of solid conclusions, especially in the event that classification
and normalization is not performed. In addition to the classification
and normalization, a clear need is recognized so as to propose and
standardize new global indicators expressing properly the specific
energy penalty.
The present paper aims to investigate in depth existing knowl-
edge on the energy impact of urban warming, the latter as caused
by the urban heat island the global warming. To this end, results

http://dx.doi.org/10.1016/j.enbuild.2014.07.022
0378-7788/© 2014 Elsevier B.V. All rights reserved.
 M. Santamouris / Energy and Buildings 82 (2014) 100–113
as published in several articles are collected, classified and ana-
lyzed; in addition new energy impact indicators are defined and
calculated.
2. Estimating the energy impact of urban warming
The impact of high ambient temperatures on the energy con-
sumption and the environmental quality of buildings is a quite
well investigated area. Recent research mainly focus on two dif-
ferent categories of studies aiming to evaluate the energy impact
of urban heat island and global warming:
(a) Studies aiming to evaluate the energy impact of the urban heat
island. These studies rely on the use of climatic data as collected
from urban and rural or suburban stations in order to evalu-
ate in a comparative way the precise energy needs of reference
buildings.
(b) Studies aiming to evaluate the temporal evolution of the energy
penalty to buildings as imposed by the global urban warm-
ing. These studies are mainly based on the use of long series
of past and current climatic data as collected from reference
meteorological stations.
For both categories of studies, two types of specific method-
ologies are proposed and used. The first methodology focuses on
the evaluation of the specific energy impact calculated for one or
more reference buildings, like a small or a large office, a typical res-
idence, etc. In this case, detailed or simplified energy simulation
techniques are used to calculate the energy load of the building.
The second methodology attempts to estimate the energy impact
implied on the whole building stock of a specific geographical zone.
This is performed by using methods like statistical energy model-
ing or detailed mapping and simulation of the building stock. As a
result of the various categories of studies and types of methodolo-
gies, climatic related energy impact studies for buildings involve
four ‘groups’ of investigations classified according to the climatic
problem under investigation and the evaluation methodology used.
In the following chapters the main results of existing studies on the
energy impact of high urban temperatures on buildings are ana-
lyzed on the basis of the classification described above, and are
presented accordingly.
2.1. The energy impact of heat island on various types of
buildings using urban and rural climatic data
Identification of the heat island characteristics in an urban area
requires knowledge of the spatial distribution of the ambient tem-
perature. In most of the cases, a climatic station usually located in
a thermally undisturbed suburban or rural area is selected as the
reference one. The maximum difference between the urban and
the reference station is known as the heat island intensity and is
used to characterize the local heating and overheating conditions.
Although, detailed heat island studies are available for a wide num-
ber of cities around the world, energy impact studies using urban
and rural climatic data are quite limited. Thirteen studies are pub-
lished on the topic concerning five cities in Europe (Athens, London,
Munich, Rome, Volos), four US cities and states (Boston, New York,
California, Texas), and two other cities in the rest of the world, Mel-
bourne Australia and Bahrain. Details of all studies and the main
results are given in Table 1. In most of the cases, ten, the compar-
ative analysis was performed for reference residential buildings
while the rest of the cases referred to offices. As it concerns the
number of climatic stations used, only three studies were based
on data from just two stations, the reference and an urban one,
five studies used data from three to five stations while six studies
101
used climatic inputs from more than twenty stations. The mag-
nitude of the warming in the considered cities is not reported in
the same way in all studies and comparisons are quite difficult.
Some studies give the average maximum UHI intensity as mea-
sured in summer while many others report the average annual
temperature difference between the urban and reference stations.
Reported average maximum summer urban heat island intensities
range between 3.5 K and 7 K while the annual average tempera-
tures differences were between 1 K and 3 K. To evaluate the energy
needs of the buildings, eight studies employed dynamic simula-
tion techniques, three studies were based on empirical parametric
models and two studies calculated the energy needs using sim-
ply the corresponding degree days. Three studies calculated the
energy load of the buildings without considering the efficiency of
the heating and cooling systems, three studies calculated the neces-
sary primary energy while the rest calculated the delivered energy.
Most of the studies, nine, calculated the energy impact of UHI dur-
ing both the winter and summer periods, heating and cooling load,
three studies concentrated just on the cooling impact and one study
estimated the impact on the heating demand. In parallel, two stud-
ies estimated the increase of the peak electricity demand during
the summer period and another study investigated the possible
decrease of the COP (coefficient of performance), of air conditioners
caused by warming. Details and results of all studies are presented
below.
2.1.1. Presentation of the cases under consideration
Heat island characteristics in Athens Greece are very well stud-
ied. Almost 30 stations are placed in urban and suburban locations
in and around the city measuring ambient temperature and allow-
ing the estimation of its spatial distribution. According to [13],
the average maximum intensity of heat island during the summer
period is around to 6–8 K. High urban temperatures are observed
in the central and western parts of the city characterized by
high urban density, increased anthropogenic heat and consider-
able absence of green areas. The energy impact of the heat island
phenomenon is studied in many ways. Santamouris et al. [13], has
used hourly data from 30 climatic stations and calculated the spa-
tial distribution of the heating and cooling load for a typical office
building during the summer and winter period of 1996. As it con-
cerns the cooling load it was found that the monthly load in the
center of the city was about 120% higher than the respective load
at the reference zone, while the corresponding heating load in the
center was reduced by 38%. As it concerns the peak electricity load
for cooling purposes, it increased from 13.7 kW/m2 in the reference
areas to 27.5 kW/m2 in the center of the city. Finally, the heat island
caused a serious reduction of the minimum COP values of the air
conditioning systems. In particular, while the minimum COP of the
air conditioners in cooling mode was 102% in the reference station,
it decreased to 75% in the areas associated to the highest intensity
of heat island.
In another study presented by Hassid et al. [14], the cooling as
well as the peak electricity demand of a residential building with
four apartments was calculated using climatic data from four differ-
ent stations located in Western Athens where heat island is usually
established, and two other reference stations located in suburban
zones of the city. Data for 1997 and 1998 were used. For 1997, it
was found that the cooling demand of the residential building in
western Athens was 41 kWh/m2 /y, while in the reference station
was almost 70% lower. In parallel, the peak electricity demand for
cooling in western Athens was calculated close to 30.6 W/m2 /y,
almost 100% higher than in the reference zone. For 1998, a signif-
icantly warmer year than 1997, the electricity demand was much
higher in all zones of the city and the corresponding differences
were reduced. In particular, the cooling demand in western Athens
was 45.4 kWh/m2 /y almost 29% higher than in the reference area. In
102 M. Santamouris / Energy and Buildings 82 (2014) 100–113

Table 1
Characteristics of the heat island energy impact studies using urban and rural data.

No. City, Intensity of the Characteristics of the Type of energy studies Main results Refs.
country heat island (C) used climatic data performed

1 Athens, Average Data of 30 urban and Dynamic simulation of – Increase of the cooling load by 120% in [13]
Greece maximum: sub urban stations for the heating and cooling the center of the city compared to the
6–8 K the summer of 2006 load of a typical office reference zone.
building – Decrease of the heating load in the center
by 38% compared to the reference zone
– Increase of the peak electricity demand
for cooling in the center by 100% compared
to the reference zone
– Decrease of the minimum COP of air
conditioners for by 25%
2 Western For the selected Data for 1997 and 1998 Dynamic simulation of – For 1997, the cooling load in Western [14]
Athens, stations the of 4 stations in the cooling load of a Athens for a set point temperature of 26 ◦ C
Greece average Western Athens, plus residential building was 41 kWh/m2 /y, about 67% higher than
maximum heat two reference stations with four apartments in the reference station. For 1998, the load
island intensity has increased in both areas and it was
varied between almost 29% higher in western Athens
4 and 8 K for (45.4 kWh/m2 /y) than in the reference
1997 and station
between 3 and – The peak electricity demand for cooling it
5 K for 1998 was 30.6 W/m2 in Western Athens for
1997, almost 100% higher than in the
reference zone. For 1998, the peak demand
has increased to 38 W/m2 , almost 30%
higher than in the reference
3 Volos, Average: 3.4 K Data of an urban and Dynamic simulation of – Heating load: the heating load during the [15]
Greece during winter one suburban station a building night period was 67.7 kWh/m2 /y in the
nights and suburban area and 58.7 kWh/m2 /y in the
3.1 K during central part of the city, i.e. a reduction of
summer nights 13%
– The night cooling load was 2.1 kWh/m2 /y
in the center and 1.1 kWh/m2 /y in the
reference area
4 Rome, Italy Maximum: Data from five stations Dynamic simulation of The maximum increase of the cooling load [16]
4.5 K distributed in urban the cooling load of a in the urban areas was by 130% higher than
and suburban areas typical apartment in the reference station. The load in the
during the summer urban zones varied between 18 and
period (part of July and 24 kWh/m2 . While in the reference station
August) it was close to 10 kWh/m2
5 London, UK Average annual Data from 24 different Dynamic simulation of The urban cooling load is up to 25% higher [17]
temperature climatic stations in the heating and cooling than the rural load over the year, and the
difference: 2 K London loads of a typical air annual heating load is reduced by 22%. In
conditioned office particular the lower cooling load was
positioned at 24 23.6 kWh/m2 /y and the highest
different locations in 30.3 kWh/m2 /y. The highest heating load
London was 11 kWh/m2 /y and the minimum
9 kWh/m2 /y
6 London, UK Average annual Data from 20 different Dynamic simulation of – The cooling load increased from 27 to [18]
temperature weather stations in the heating and cooling 45% between the rural and the urban areas
the difference London load of a typical office depending on the considered
is at 2 K building with different characteristics. The cooling in the central
energy conservation area varied between 24 and 43 kWh/m2 /y.
and construction – The heating load decreased by 64% in the
patterns central area compared to the rural one. The
heating load in the central area was close
to 13–14 kWh/m2 /y.
7 Munich, For 1982, Data from an urban Simple estimation of The heating load at the center of the city [20]
Germany heating degree and a rural station the heating needs of a was 17% lower than the load in the rural
hours are by residential building area
17% higher in
the city center
than in an
urban area
8 Boston, 1.3–2.8 K Data from one urban Dynamic simulation of For the small office the cooling load at the [21]
USA and two rural weather a small office and a center was 20–37% higher than in the rural
stations single family building areas. The cooling load in the rural area was
close to 5 kWh/m2 /y. The corresponding
heating load in the urban area was
38 kWh/m2 /y and it was 2–10% lower than
in the corresponding rural areas.
For the single family house, the cooling
load in the urban area was 3 kWh/m2 /y
and it was almost 3–9% higher than in the
rural areas. The heating load in the urban
area was 82 kWh/m2 /y and it was 14%
lower than in the rural areas
 M. Santamouris / Energy and Buildings 82 (2014) 100–113 103

Table 1 (Continued)

No. City, Intensity of the Characteristics of the Type of energy studies Main results Refs.
country heat island (C) used climatic data performed

9 State of Based on the Data from surface as Calculation of the Cooling load: the primary energy of a [22]
New York, mean well as from the NNRa heating and cooling household not influenced by the UHI varied
USA temperature stations, load of buildings using between 89 and 209 kWh/y. Because of the
data it varies a parametrical model. heat island it increased to 286–683 kWh/y,
between 0 and Calculations are i.e. an increase close to 220%.
2K performed for both Heating load: the primary energy for a
climatic data sets. The household not influenced by the UHI
difference is the energy varied between 10,908 and 33,151 kWh/y.
due to the UHI Because of the heat island it decreased to
10,194–30,980 kWh/y, i.e., a decrease of
about 7%.
Total load: the UHI decreased the annual
total primary energy spent for heating and
cooling by 5%
10 State of Based on the Data from surface as As above Cooling load: the primary energy of a [22]
California, mean well as from the NNRa household not influenced by the UHI
USA temperature stations varied between 623 and 831 kWh/y.
data it varies Because of the heat island it increased to
between 0 and 1994–2658 kWh/y, i.e. an increase close to
4K 220%.
Heating load: the primary energy for a
household not influenced by the UHI
varied between 4220 and 6618 kWh/y.
Because of the heat island it decreased to
2721–4267 kWh/y, i.e., a decrease of about
55%.
Total load: the UHI decreased the annual
total primary energy spent for heating and
cooling by 3–7%
11 State of Based on the Data from surface as As above Cooling load: the primary energy of a [22]
Texas, USA mean well as from the NNRa household not influenced by the UHI
temperature stations varied between 9167 and 17,157 kWh/y.
data it varies Because of the heat island it increased to
between 0 and 9686–18022 kWh/y, i.e. an increase close
1K to 6%.
Heating load: the primary energy for a
household not influenced by the UHI
varied between 3111 and 5169 kWh/y.
Because of the heat island it decreased to
2680–4453 kWh/y, i.e., a decrease of about
16%.
Total load: the UHI decreased the annual
total primary energy spent for heating and
cooling by 1%
12 Melbourne, 1.4 K mean Data from one urban Calculation of the For the existing building the cooling load [23,24]
Australia annual and one rural weather heating and cooling in the central areas was 8 kWh/m2 /y, 10%
difference station load of a typical higher than in the rural area. For the new
existing and a new building, the cooling load in the central
residential building area was 5 kWh/m2 /y and was 17% higher
than in the rural area.
The heating load in the central area was
97 kWh/m2 /y and 24 kWh/m2 /y for the
existing and the new building respectively
and was 5% and 7% lower compared to the
corresponding rural values
13 Bahrain Average Climatic data from five Analysis of real energy The average annual increase of the cooling [25]
maximum heat weather stations consumption data from load in the urban area was 18%. The
island intensity around and in the city a quite high number of minimum annual cooling in the less urban
close to 3 K residential buildings zone was close to 72 kWh/m2 /y
a
NNR = (NCEP–NCAR (National Centers for Environmental Prediction/National Center of Atmospheric Research 50-year Reanalysis).

parallel, the peak electricity demand for cooling increased to
38 W/m2 , that is, 30% higher than in the reference area. The
impact of the set point temperature on the magnitude of the
observed cooling differences was investigated; the higher the set
point temperature the higher the difference in percentage. In
particular for 1997 and for set points of 24 ◦ C, 25 ◦ C, 26 ◦ C and
27 ◦ C, the corresponding differences were 55%, 61%, 70% and 81%.
This is because the ratio of the cooling degrees hours in the
two stations, as calculated for the given set point temperature,
increased as the set point rises. However, when ambient temper-
atures increase in the whole city because of the given synoptic
conditions, for example during a heat wave, the values of the
absolute cooling demand are increasing everywhere, however the
intensity of heat island is significantly reduced and the correspond-
ing cooling load differences decrease. For example in 1998, and
for the same set point temperatures (24–27 ◦ C), the corresponding
differences of the cooling demand were 24, 26, 28 and 31%. The
energy penalty of the increased ventilation rates applied in build-
ings affected by the urban heat island, was found to be important. In
particular, for 1997, while an increase of the ventilation rate from
0.5 ach to 2 ach increased the cooling demand by 3% in the reference
station, the corresponding increase in western Athens was close to
104 M. Santamouris / Energy and Buildings 82 (2014) 100–113
25%. However, when synoptic conditions favor higher urban tem-
peratures, the penalty of the increased ventilation rates increases
faster in the reference zones than in the UHI affected areas. In par-
ticular, for a similar increase of the ventilation rate, the cooling
penalty in the reference station increased from 3% to 20% while in
the UHI affected zones the corresponding increase was from 25% to
27%. This is because during heat waves, night time temperatures are
increasing considerably in the whole city, thus the cooling potential
of ventilation in the reference zones is greatly reduced.
The energy impact of night time heat island developed in the
medium size city of Volos in Greece is studied in [15]. Using data
from an urban and suburban climatic station, the heating and cool-
ing demand of a typical building is simulated for a complete year.
As reported, heat island reduces the night time heating demand in
the urban area by 13%. Climatic data from five urban and suburban
stations located in Rome Italy, were used to estimate the cooling
energy penalty due to the urban heat island in the city [16]. The
study refers to the summer period and in particular for the days
between 18th of July and 7th of September. The maximum intensity
of the heat island was measured close to 4.5 K. The cooling load of a
typical apartment was calculated using dynamic simulation tech-
niques. It was found that because of the heat island phenomenon,
the cooling load in the urban areas could increase up to 130%.
Heat island and its impact on the energy demand of build-
ings is very well studied for London, UK. Chandler [17] has used
climatic data from 1951 to 1960 and found that heating degree
days in central London were reduced to about 10% compared to
the surrounding rural areas. Kolokotroni et al. [18] have calcu-
lated the heating and cooling demand of a typical air conditioned
office building using data from twenty four climatic stations located
around the greater London area. The main result of the simulation
study was that the cooling demand in the urban areas increased
by 25% compared to the rural areas while the heating demand
was reduced in the central London areas by 22%. The study calcu-
lated much higher cooling than heating demands as a result of the
increased internal gains of the considered buildings. The specific
results presented a high variability as a function of the considered
internal gains and the layout of the zone where the building was
located. In another study aiming also to identify the energy penalty
of the heat island phenomenon in London [19], the heating and
cooling demand of a typical office building was calculated using
climatic data from twenty weather stations located in and around
London. It was found that heat island increases the cooling demand
of the building in the center of the city by 27–45% depending on
the specific energy conservation features considered. In parallel,
the heating demand in the central area was almost 64% lower than
in the surrounding rural areas.
The potential decrease of the heating needs of a residential
building in Munich, Germany because of the heat island phe-
nomenon was investigated in [20]. Data for 1982 measured in a
rural and an urban station were used. It is found that the heating
degree days and the corresponding heating load decreases by 17%
in the urban areas compared to the corresponding load in the rural
zone.
The energy impact of urban heat island in Boston, USA, was eval-
uated in [21]. As mentioned, the measured heat island intensity
varied between 1.3 and 2.8 K depending on the selected reference
station. Climatic data from an urban and two rural weather stations
were used to simulate the energy performance of a single family
residential building and a small office. For the small office it was
calculated that the heat island decreased the heating demand in
the central area of the city by 2–10% depending on the used refer-
ence station. The corresponding decrease of the heating demand
for the single family house was 14%. As it concerns the cooling
demand, it was found to be very small for both the office and the
single family buildings while the percentage increase in the urban
area was 20–37% for the small office and 3–9% for the residential
building.
An analysis on the energy consequences of the heat island in
three States of USA, New York, California and Texas is given in [22].
Climatic data from surface stations, as well as from the NCEP–NCAR
(National Centers for Environmental Prediction – National Center
of Atmospheric Research) 50-year Reanalysis, NNR stations were
used. It is considered that the NNR stations are not influenced by
the urban heat island phenomenon and thus the difference between
the two data sets reveals the climatic impact of the phenomenon in
the given place. Parametric models to predict the primary energy
used for the heating and cooling of the buildings in the three con-
sidered states were developed and used. As it concerns the State
of New York, it was calculated that the heating and cooling degree
days given the NNR climatic data were 7298 and 217 respectively,
while the corresponding values given by the surface stations were
6820 and 315. Thus, it is considered that the UHI phenomenon has
added 98 additional CDD and decreased the HDD by 428. This corre-
sponded to an additional primary cooling demand of 89–210 kWh/y
per household and a reduction of the primary heating demand by
714–2170 kWh/y/household. In total, it is reported that because of
the urban heat island phenomenon, the total energy consumption
of a household in the NY state decreased by 3–7%.
For the State of California, climatic calculations showed that the
urban heat island increased the cooling degree days from 394 to
1260 and decreased the heating degree days from 8016 to 5088.
This resulted in an increase of the primary cooling demand of a
household by 1371–1827 kWh/y and a decrease of the correspond-
ing heating demand by 1500–2351 kWh/y. In total, it is reported
that the urban heat island in California decreased the global energy
consumption of a household by 3–7%.
Finally, for the State of Texas, an increase of the cooling degree
hours was observed from 2564 to 2709 and a decrease of the
heating degree hours from 2435 to 1464 This corresponded to an
increase of the primary cooling demand by 519–965 kWh/y per
household and an increase of the corresponding heating demand
by 431–716 kWh/y. In total, urban heat island in the state of Texas
increases the global energy consumption of households by 1%.
An estimation of the energy impact of the urban heat island in
Melbourne, Australia is given in [23,24]. Climatic data from one
urban and one rural station was used. The average annual heat
island intensity was found to be close to 1.4 K. Dynamic simulations
were performed to calculate the heating and cooling demand of an
existing as well as of a new building. It was found that because of the
heat island, the cooling demand of an existing building increased
by 10% while the corresponding load for a new building increased
by 17%. As it concerns the heating demand, it was calculated that
in the central area and for the existing and new buildings was 5%
and 7% lower than the corresponding values in the rural zone.
Studies on the impact of urban heat island in Bahrain are
reported in [25]. Data from five weather stations distributed around
the city were used and the average maximum heat island inten-
sity was reported to be close to 3 K. It was also found that the
annual cooling degree hours for the urban areas were about 17%
higher than in the suburban areas. Electricity consumption data
from a high number of residential buildings were collected, while
regression based models were used to predict the cooling energy
consumption using the cooling degree days as input. The cooling
load of the residential buildings in the more warm urban areas was
found to be 18% higher than in the less warm area of the city.
Based on the above results the following conclusions can be
drawn:
(a) In cooling dominated zones, the increase of the cooling demand
triggered by the urban heat island is much higher than the
corresponding decrease of the heating demand for all types of
 M. Santamouris / Energy and Buildings 82 (2014) 100–113
Fig. 1. Relation of the cooling energy demand of individual buildings as calculated for the reference and the urban stations.
studied buildings. On the contrary, in heating dominated zones,
urban heat island results in a significant decrease of the total
energy demand of residential buildings as the heating load is
reduced. As shown in [22], a rise of the ambient temperature
may not increase the cooling load in zones with summer tem-
perature below 23 ◦ C, but will increase the cooling load in areas
with average summer temperature above 27 ◦ C. However, in
tertiary buildings, the impact of higher ambient temperatures
on the total energy demand strongly depends on the amount of
internal thermal gains. As shown in the London’s study [18], in
offices with high internal gains the cooling demand is dominant
and a potential increase of the ambient temperature results
in a higher total energy consumption. On the contrary when
the internal gains are reduced, heating is the dominant energy
component and increased UHI intensities help to decrease the
total energy demand. Fig. 1 compiles all available information
regarding the calculated increase of the cooling energy con-
sumption of individual buildings caused by the urban warming.
In particular, the cooling energy demand of individual build-
ings as calculated for the reference station is plotted against the
corresponding cooling demand for the urban area station. The
figure includes the data for all types of buildings and geographic
zones presented in the above analysis. As shown, the average
increase of the cooling load caused by the UHI is close to 13.1%
compared to the corresponding load calculated using the data
of the rural reference stations (reference demand). The corre-
lation coefficient R2 was quite high, equal to 0.97. It is evident
that the above figure is indicative and is based on a satisfactory
but quite limited number of studies. When more studies are
available and additional data are added into the data base, the
corresponding figure may improve.
(b) Given that the cooling penalty caused by the UHI is a function
of the warming characteristics of the given location and also
of the specific building features, the possible relations between
these three parameters are investigated by using the data of
the previously presented thirteen studies. It is found that the
calculated annual cooling penalty per degree of average UHI
intensity, as measured during the reporting period, is very well
105
correlated against the logarithm of corresponding reference
cooling demand (Fig. 2). As expected, the higher the reference
cooling demand, the higher the absolute value of the cool-
ing penalty per degree of UHI intensity, however the relation
between the two parameters is logarithmic and not linear. This
happens because in buildings with high cooling demands the
envelope related convective and conductive part of the load
which is influenced by the presence of the UHI is proportionally
reduced compared to the internal and solar thermal gains.
(c) Under hot weather synoptic atmospheric conditions the energy
impact of the urban heat island is seriously reduced. Previously
published research has shown that under the specific synoptic
conditions the day time intensity of the heat island is reduced, as
temperature differences are minimized in the city [4,26]. This
results in decreased cooling energy consumption differences
between the urban and reference zones although the absolute
cooling energy demands are increased. This is evident from the
analysis of the 1997 ad 1998 data in Athens [14], where a seri-
ous decrease of the UHI energy impact is calculated under hot
weather synoptic conditions. Contrary, to the above, computer
simulations have shown that under the specific conditions, the
night time heat island intensity may be slightly higher because
of the increased thermal storage in the zones of high den-
sity [27]. However, the expected energy impact on the cooling
demand is almost negligible.
(d) The proportional variation of the cooling demand in cities
caused by the urban heat island is a function of many
design and operational characteristics of the considered
buildings. The set point temperature, ventilation rate and
ventilation type and the insulation levels are among the
more decisive parameters. When the set point temperature
increases, the absolute cooling demand decreasing decreases;
however, the difference between the reference cooling
load and the one calculated for the urban area increases well.
This is because the ratio of the time that the ambient tempera-
ture exceeds the set point in the urban areas against the corre-
sponding time in the reference zones is increasing as a function
of the selected base temperature. The difference tends to a
106 M. Santamouris / Energy and Buildings 82 (2014) 100–113
Fig. 2. Correlation between the references cooling demand of a building against the corresponding cooling penalty per degree of UHI intensity.
minimum when hot weather synoptic conditions are pre-
vailing in the whole area. Applying high ventilation rates in
zones affected by the urban heat island results in an impor-
tant increase of the cooling demand compared to the reference
areas. This is an obvious result as the supply of the building with
higher temperature fresh air requires an increased consump-
tion to cool it. In parallel, an important increase of the night
time urban heat island minimizes the ability of night ventilative
cooling techniques to precool the building and thus increases
the total cooling demand. Higher insulation levels in a build-
ing reduce significantly the losses through the envelope and
contribute to lower heating loads and slightly higher cooling
demands. Well insulated buildings in cities having an important
heat island problem are better protected than buildings with-
out or with limited insulation and present a reduced cooling
penalty regarding the reference cooling consumption.
(e) It is evident that the cooling penalty triggered by the urban
heat island is a strong function of the temperature differences
between the reference and the urban stations. When a full set of
climatic data is available, cooling degree days may be computed
for the reference and the urban stations. When the cooling
load of the building is not highly influenced by solar and inter-
nal thermal gains, the cooling demand of a building is mostly
determined by the convective and conductive gains and losses
through the envelope and thus cooling degree days may be used
to estimate the specific energy demand of a building and the
corresponding impact of the urban heat island. However, such a
methodology presents some serious limitations and may be the
source of significant inaccuracies for buildings with important
radiative thermal gains or internal sources. Therefore, studies
reporting the cooling impact of UHI using evaluation methods
which link in a linear way the cooling degree days and the spe-
cific energy consumption of a building should be considered
with extreme care.
2.2. The energy impact of UHI and climate change on the total
building stock of a city
Calculation of the energy penalty for a single building is impor-
tant because it provides a useful and rapid indication of the impact
of urban warming. However, the specific results depend fully on
the characteristics of the selected building and they are not repre-
sentative for the whole building stock of a city. To overcome the
problem, several studies have attempted to estimate the energy
impact of urban warming of the global building stock of a city. Cal-
culations of the global energy penalty in a city are performed using
the two methodologies.
The first methodology uses a two steps approach. At a first step,
the cooling energy consumption of typical buildings is calculated.
The total energy demand in the city is estimated in the second step
using as inputs the typical consumptions for each building type
and also the size, distribution and statistical characteristics of the
whole building stock in the city. The above approach requires two
sets of climatic data. The first data set includes actual temperature
measurements from the concerned location as affected by the UHI,
and the second one involves measured temperature data from one
or more reference stations. During the first step, the cooling energy
consumption of all typical buildings is calculated using both data
sets. Then, projections for all buildings are performed using the
actual knowledge on the size, characteristics and the distribution
of the whole building stock. The difference between the two esti-
mations is considered as the cooling penalty triggered by the UHI.
This methodology is used in [12,14], to evaluate the global cool-
ing impact of UHI in Athens Municipality and the whole region of
Western Athens.
The second methodology aims to establish correlations between
the ambient temperature in the city and the corresponding energy
consumption for cooling purposes. Using simulation or empiri-
cal methods two sets of climatic data are created: one where the
impact of UHI is included and a second one where it is neglected.
Correlations between the ambient temperature and the cooling
energy consumption are developed and used for both data sets and
the corresponding total energy consumption is estimated. The dif-
ference between the two calculations is considered as the penalty
caused by the UHI. This second methodology is used in Tokyo [28]
and Beijing [29].
Following the characteristics of the studies aiming to predict the
global energy penalty of UHI and climate change in Athens, Western
Athens, Tokyo and Beijing are presented and discussed. Details of
the existing studies as well as their main characteristics and results
are given in Table 2.
 M. Santamouris / Energy and Buildings 82 (2014) 100–113 107

Table 2
Characteristics of the studies aiming to predict the global energy penalty of UHI and climate change.

No. City, country Heat island Characteristics of the Type of energy analysis Main results Refs.
intensity used climatic data

1 Western Average Data for 1997 and 1998 Calculation of the global final Final cooling demand: for 1997 the [14]
Athens, maximum: of four station in electricity demand for cooling maximum additional energy for cooling in
Greece 4–8 K for 1997 Western Athens, plus and peak electricity for western Athens was 180 MWh/y or
3–5 K for 1998 one reference station Western Athens using data 1015 kWh per dwelling and 405 kWh per
from the four local stations and local citizen or 23 kWh/m2 /y. For 1998, the
the reference one and maximum additional energy was
estimation of the additional 133 MWh/y or 750 kWh per dwelling or
energy and peak demand 300 kWh per person and
caused by the heat island 16 kWh/m2 /y kWh/m2 /y.
Peak electricity: for 1997 the maximum
additional peak electricity demand was
334 MW or 1.9 kW per dwelling and 0.7 kW
per person, while for 1998 the maximum
additional peak electricity demand was
168 MW or 0.95 kW per dwelling, 0.38 kW
per person
2 Athens, The average Data for 1997 and 1998 Dynamic simulation of the Considering that all buildings are [12]
Municipality, maximum of 4 stations located in cooling load of different types air-conditioned, for 1997 the maximum
Greece intensity varied the Municipality of of buildings for all stations additional cooling energy needs for the
between Athens and one whole Municipality of Athens was
4–9.5 K for reference station 1540 GWh or 33.2 kWh/m2 /y For 1998, the
1997, and corresponding values were 1346 GWh and
4–8 K for 1998 29 kWh/m2 /y. The actual energy penalty
for 1997 and 1998 was equal to 120.7 GWh
and 107.5 GWh respectively
3 Tokyo, Japan Average: 2.5 K Simulation of the Development of hourly The cooling penalty for the whole city was [28]
in summer and temperature patterns correlations between the 3444 GWh/y. The cooling penalty for the
1.5 K in winter in the whole Tokyo heating and cooling energy commercial sector was 2360 GWh/y and
area consumption and the ambient for the residential sector 1741 GWh/y, or a
temperature for each district. total penalty of 384 kWh per person per
Integration over a complete year.
year using the corresponding The decrease of the heating and hot water
simulated climatic data demand in the commercial sector was
2055 GWh/y or 229 kWh/person/y, while
the corresponding decrease in the
residential sector was 3582 GWh/y or
400 kWh/person/y. In total, the global
energy consumption decreased by 3.7% per
year
4 Beijing, China Average: 4–6 ◦ C Climatic data of Beijing Development of empirical The calculated additional energy spent for [29]
for 2005 correlations between the air condition in the Beijing area for 2005 is
energy consumption for 815,299 MW h plus 287,455 MW h and
cooling and temperature 217,980 MW h for reasons related to the
urban overheating

The required additional final electricity consumption and peak
electricity demand because of the urban heat island phenomenon
was estimated for the whole area of Western Athens, Greece, in
[14]. Data from four local climatic stations collected during 1997
and 1998 were used together with a reference station. The whole
area involves 177,385 dwellings with almost 438,000 inhabitants.
Dynamic simulations of the energy consumption of different types
of buildings were performed using the previous sets of climatic
inputs. Using data on the existing stock, the cooling energy con-
sumption as well as the peak electricity demand was calculated for
both the local and the reference weather conditions. It was esti-
mated that for 1997, the maximum additional energy for cooling
was 180 MWh/y which corresponds to 1015 kWh per dwelling and
405 kWh per citizen. For 1998, the corresponding maximum addi-
tional cooling load was 133 MWh/y or 750 kWh per dwelling or
300 kWh per person. In parallel, the maximum additional peak elec-
tricity demand caused by the urban heat island was estimated close
to 334 MW and 168 MW for 1997 and 1998 respectively. This cor-
responds to an additional peak demand of 1.9 kW and 0.95 kW per
dwelling and 0.7 kW and 0.38 kW per person for 1997 and 1998
respectively.
A similar study aiming to investigate the global cooling energy
penalty for the Municipality of Athens, Greece is given in [12]. Data
for 1997 and 1998, from four urban stations located inside the
borders of the Municipality as well from a reference station were
used. Dynamic simulation techniques were used to calculate the
cooling needs of different types of buildings and then the sum-
mer energy penalty caused by the urban heat island was calculated.
Using available data on the distribution and the size of the building’s
stock in the city, it was calculated that the global cooling penalty for
1997 and 1998 were close to 1540 GWh/y and 1346 GWh/y respec-
tively. The above numbers correspond to about 33.2 kWh/m2 /y and
29 kWh/m2 /y for the two considered years respectively. Compared
to the values reported previously for Western Athens, the penalty
per square meter of building in the Municipality of Athens seemed
to be 40–80% higher. This is because most of the area of the Munic-
ipality of Athens experiences high urban heat island intensities
while in Western Athens the phenomenon presents a high inten-
sity in specific areas while the intensity in the rest of the region is
moderate.
A detailed study aiming to evaluate the energy impact of the
urban heat island in Tokyo, Japan is described in [28]. The specific
temperature distribution in the Tokyo area was simulated using
a mesoscale climatic model. Simulations were performed for the
actual situation as well as for a scenario that the climatic impact of
heat island is not taken into account. The calculated average heat
island intensity was 2.5 K in summer and 1.5 K in winter. Using
actual energy consumption data, correlations between the hourly
108 M. Santamouris / Energy and Buildings 82 (2014) 100–113
energy consumption and the corresponding ambient temperature
were obtained and used to estimate the specific energy consump-
tion for heating and cooling for the whole building stock for the
two data sets both considering and neglecting the climatic impact
of heat island. Separate estimations were performed for the com-
mercial and the residential sectors. In particular, the calculated
cooling penalties for the commercial and residential sectors were
2360 GWh/y and 1741 GWh/y. Considering that the population in
the 23 wards of Tokyo is approximately 8,950,000 persons, the
total penalty per person was estimated close to 384 kWh, out of
which 263 kWh were allocated to the residential sector. In parallel,
the total decrease of the energy consumption for heating and hot
water was estimated close to 2055 GWh/y for the commercial sec-
tor and 3582 GWh/y for the residential sector. As a result, the global
energy impact of the urban heat island in Tokyo was equivalent
to a decrease of the total energy by 3.7% per year, while the con-
sumption on the commercial sector increased by 1% per year and
in the residential sector decreased by 8%. As it concerns the spe-
cific energy use, the urban heat island increased the cooling energy
consumption by 27.5%, and decreased the heating and hot water
consumption by 18.4% and 6.8% respectively.
The additional energy spent for air conditioning because of the
urban heat island in Beijing, is analyzed in [29]. The study was
performed for 2005 and refers to the central zone of the city. The
intensity of the urban heat island in the city was measured between
4 and 6 K [30]. Two empirical relations between the energy pro-
duced for air condition and the urban temperature were developed.
The first relation used temperature data considering the influence
of the local urban heat island while the second used undisturbed
temperature data. The difference between the energy consump-
tion calculated by the two relations was defined as the additional
energy spent for air conditioning. The annual direct additional
energy because of the urban warming was estimated equal to
1,320,664 MWh. This means that heat island and global warming
increases the annual energy spent for cooling in the city by almost
48%.
Based on the above data, four homogenized indicators are pro-
posed and calculated representing the actual global penalty of
urban heat island and climate change in a city. In particular, the
proposed indicators are:
The global energy penalty per unit of city surface (kWh/m2 ).
The global energy penalty per unit of city surface and per degree
of the UHI intensity (kWh/m2 /K).
The global energy penalty per person (kWh/p).
The global energy penalty per person and per degree of the UHI
intensity (kWh/p/K).
whereas results are illustrated in Fig. 3.
The global energy penalty per unit of city surface (GEPS), is an
indicator highly influenced by the density of buildings in the city;
the higher the density of buildings per square kilometer in the
city the higher the value of the indicator for a given heat island
intensity, building characteristics and air condition penetration
levels. The GEPS indicator is found to vary between 1.1 kWh/m2
for Beijing and 5.5 kWh/m2 for Tokyo. The corresponding values
for the commercial and residential sector in Tokyo are 3.8 and
1.7 kWh/m2 respectively. For western Athens, the GEPS indica-
tor varies between 2.3 and 3 kWh/m2 , and for the Municipality of
Athens between 1.2 and 1.4 kWh/m2 . Although the UHI intensity in
Athens and western Athens is quite higher than the specific zone
of Tokyo, the GEPS index values in Athens are significantly lower.
This is because the buildings density is much lower in Athens than
in Tokyo while the penetration of air conditioning is also much
lower.
The indictor related to global energy penalty per unit of city sur-
face and per degree of the average UHI intensity, GEPSI, reflects
the same characteristics as the GEPS indicator by taking into
account the average UHI intensity characteristics in the considered
city. Values of GEPSI, vary between 2.2 kWh/m2 /K for Tokyo and
0.24 kWh/m2 /K for the Municipality of Athens. The corresponding
value for Beijing was 0.42 kWh/m2 /K while for western Athens was
between 0.56 and 0.76 kWh/m2 /K.
The global energy penalty per person, GEPP, is an indicator
highly influenced by the density of people in the area concerned.
The lower the population the higher the value of the GEPP for a
given heat island intensity, building characteristics and air condi-
tion penetration levels. The GEPP index is found to vary between
104 kWh for Athens and 405 kWh for western Athens. Beijing has
a GEPP value close to 113 kWh while Tokyo is around to 385 kWh.
Higher values are calculated for western Athens as it is the area
with the lowest population density among the existing studies and
also presents the highest urban heat island intensity.
Finally, the global energy penalty per person and per degree
of the UHI intensity, GEPPI, has the same characteristics as the
GEPP indicator while it includes the local UHI intensity as addi-
tional information. Values of GEPPI varied between 20 kWh/K for
the Municipality of Athens to 154 kWh/K for Tokyo. The corre-
sponding value for Beijing was 45 kWh/K while for western Athens
it varied between 75 and 101 kWh/K.
Based on the limited available data and studies it may be con-
cluded that UHI triggers an average energy penalty per unit of city
surface, GEPS, close to 2.4 (±1.5) kWh/m2 , a global energy penalty
per unit of city surface and per degree of the UHI intensity, GEPSI,
close to 0.74 (±0.67) kWh/m2 /K, a global energy penalty per per-
son, GEPP, close to 237 (±130) kWh/p and a global energy penalty
per person and per degree of the UHI intensity, GEPPI, close to 70
(±45) kWh/p/K. The above numbers are based on a limited number
of studies and may change when more relevant data are available.
However, the given values are quite indicative of the global energy
penalty induced by urban heat island and may be used as the basis
to built additional studies and estimations.
3. The multiyear annual variation of the urban warming on
the energy consumption of various types of buildings
Multiyear ambient temperature records are available for many
places around the world. Analysis of the existing time series reveals
important warming tendencies for most of the places. Average
warming trends may be as high as 1.0 K per decade like in Athens,
Greece [31], or may be completely insignificant as in Hong Kong
[32]. In many places warming trends are much higher during the
summer than the winter and this causes a much larger increase
of the cooling load compared to the corresponding decrease of the
heating load of buildings.
Time series of the ambient temperature are used in eighteen
cities around the world to calculate the evolution of the heating
and cooling load of representative buildings throughout the years.
The concerned cities are Athens, Larisa, Corfu and Heraklion in
Greece, Nicosia, Paphos, Limassol, Larnaka, Famagusta and Kery-
nia in Cyprus, Zurich, Geneva, Lugano and Davos in Switzerland,
Phoenix, Washington DC, Puerto Rico in US, Hong Kong in China
and Resolute in Canada. In all places, large and small offices as well
as residential buildings are considered as case studies. Except in
Cyprus and Switzerland where the energy load is calculated as a
function of the degree days, for all other places dynamic simula-
tion techniques are used to calculate the annual energy needs. For
all cases except Cyprus both the heating and cooling loads were
calculated. Details of all the considered studies as well as the main
results are given in Table 3.
 M. Santamouris / Energy and Buildings 82 (2014) 100–113
Table 3
Characteristics of the studies aiming to calculate the temporal evolution of the heating and cooling load of representative buildings.
No. City, Possible temperature increase
country
1 Athens, For the period 1970–2010:
Greece summer at
14:00:0.94 K/decade, winter at
14:00:0.19 K/decade
2 Larisa, For the period 1970–2010:
Greece summer at 14:00:0.5 K/decade,
winter: 0.16 K/decade
3 Corfu, For the period 1970–2010:
Greece summer at
14:00:0.54 K/decade, winter:
0.0 K/decade
4 Heraklion, For the period 1970–2010:
Greece summer at 14:00:
0.42 K/decade, winter:
0.2 K/decade
5 Nicosia, For the period 1900–2000 the
Cyprus temperature increase trend is
+1 K per 100 year. Between
1950 and 2010 the cooling
degree days are increased from
610 to 900
6 Pafos, Between 1950 and 2010 the
Cyprus cooling degree days are
increased from 250 to 400
7 Limassol, For the period 1900–2000 the
Cyprus temperature increase trend is
+1.3 K per 100 year. Between
1950 and 2010 the cooling
degree days are increased from
600 to 800
8 Famagusta, Between 1950 and 2010 the
Cyprus cooling degree days are
increased from 620 to 820
9 Kerynia, Between 1950 and 2010 the
Cyprus cooling degree days are
increased from 400 to 550
10 Larnaca, Between 1950 and 2010 the
Cyprus cooling degree days are
increased from 630 to 870
11 Davos, For the period 1901–2003, the
Switzerland gradient was 0.165 K/decade,
and for the period 1983–2003
it was 0.642 K/decade. For the
same period, the decrease of
the HDD was 250–270 per
decade, and the increase of the
CDD was 1.1/decade
12 Geneva, For the period 1901–2003, the
Switzerland gradient was 0.161 K/decade,
and for the period 1983–2003
it was 0.795 K/decade. For the
same period, the decrease of
the HDD 250–280 per decade,
and the increase of the CDD
was 57.6/decade
13 Zurich, For the period 1901–2003, the
Switzerland gradient was 0.143 K/decade,
and for the period 1983–2003
it was 0.776 K/decade. For the
same period, the decrease of
the HDD was 250 per decade,
and the increase of the CDD
was 42.1/decade
14 Lugano, For the period 1901–2003, the
Switzerland gradient was 0.117 K/decade,
and for the period 1983–2003
it was 0.795 K/decade. For the
same period, the decrease of
the HDD was 250–280 per
decade, and the increase of the
CDD was 65.9/decade
109
Characteristics of the Type of energy analysis Main results Refs.
climatic data used performed
40 years of hourly Dynamic simulation of Cooling load: it has increased from [31]
climatic data, the heating and cooling 99.5 kWh/m2 /y in 1970 to 124.8% in 2010
1970–2010 load of a typical office Heating load: it has decreased from
building for all 40 years 39.4 kWh/m2 /y in 1970 to 31.7 kWh/m2 /y
40 years of hourly Dynamic simulation of Cooling load: it has increased from [31]
climatic data, the heating and cooling 87.2 kWh/m2 /y in 1970 to 100.7 kWh/m2 /y
1970–2010 load of a typical office in 2010, Heating load: it has decreased
building for all 40 years from 38.2 kWh/m2 /y in 1970 to
45.5 kWh/m2 /y
40 years of hourly Dynamic simulation of Cooling load: it has increased from [31]
climatic data, the heating and cooling 81.9 kWh/m2 /y in 1970 to 101.2 kWh/m2 /y
1970–2010 load of a typical office in 2010, Heating load: it has decreased
building for all 40 years from 45.5 kWh/m2 /y in 1970 to
36.2 kWh/m2 /y
40 years of hourly Dynamic simulation of Cooling load: it has increased from [31]
climatic data, the heating and cooling 93.7 kWh/m2 /y in 1970 to 120.9 kWh/m2 /y
1970–2010 load of a typical office in 2010, Heating load: it has decreased
building for all 40 years from 34.3 kWh/m2 /y in 1970 to
26.5 kWh/m2 /y
Hourly climatic data for Calculation of the Increase of the cooling load from [33]
the period 1950–2010 cooling load of a typical 15 kWh/m2 /y in 1950 t/m2 /y to 20 kWh in
residential building 2010
Hourly climatic data for Calculation of the Increase of the cooling load from [33]
the period 1950–2010 cooling load of a typical 6.5 kWh/m2 /y in 1950 t/m2 /y to 1.5 kWh in
residential building 2010
Hourly climatic data for Calculation of the Increase of the cooling load from [33]
the period 1950–2010 cooling load of a typical 14 kWh/m2 /y in 1950 t/m2 /y to 20 kWh in
residential building 2010
Hourly climatic data for Calculation of the Increase of the cooling load from [33]
the period 1950–2010 cooling load of a typical 15 kWh/m2 /y in 1950 t/m2 /y to 20 kWh in
residential building 2010
Hourly climatic data for Calculation of the Increase of the cooling load from [33]
the period 1950–2010 cooling load of a typical 9 kWh/m2 /y in 1950 t/m2 /y to 12.5 kWh in
residential building 2010
Hourly Climatic data Calculation of the Increase of the cooling load from [33]
for the period cooling load of a typical 16 kWh/m2 /y in 1950 t/m2 /y to 22 kWh in
1950–2010 residential building 2010
Mean monthly Calculation of the Reduction of the heating load by 10% for [34]
temperature data from heating and cooling the period 1983–2003 The corresponding
1900 to 2003 load of a typical increase of the cooling load was negligible
building using the HDD
and CDD values
Mean monthly Calculation of the Reduction of the heating load between 15 [34]
temperature data from heating and cooling and 21% for the period 1983–2003. The
1900 to 2003 load of a typical corresponding increase of the cooling load
building using the HDD was close to 68%
and CDD values
Mean monthly Calculation of the Reduction of the heating load between 13 [34]
temperature data from heating and cooling and 16% for the period 1983–2003. The
1900 to 2003 load of a typical corresponding increase of the cooling load
building using the HDD was close to 35%
and CDD values
Mean monthly Calculation of the Reduction of the heating load between 18 [34]
temperature data from heating and cooling and 25% for the period 1983–2003. The
1900 to 2003 load of a typical corresponding increase of the cooling load
building using the HDD was close to 45%
and CDD values
110 M. Santamouris / Energy and Buildings 82 (2014) 100–113

Table 3 (Continued)

No. City, Possible temperature increase Characteristics of the Type of energy analysis Main results Refs.
country climatic data used performed

15 Zurich, For the period 1981–2003 the
Switzerland average annual gradient of the
temperature increase was
0.52 K/decade, while during the
summer and winter it was 0.54
and 0.53 K/decade respectively
16 Hong Kong When the data were grouped
into four 10 year periods, there
was a 0.4 ◦ C rise in the dry bulb
temperature from 1961–1970
to 1991–2000
17 Phoenix, Minimum temperature rise of
USA 0.47 ◦ C per decade from 1960
to 2000
18 Phoenix, For the urban station, the
USA annual max temperature trend
was 0.193 K/decade, while the
annual min temperature trend
was 0.984 K/decade. The values
for the rural station are:
0.03 K/decade and
0.427 K/decade for the max and
min temperatures respectively
19 Washington, The heating degree days in
DC, USA 1969 was 2970 (base 18 ◦ C),
and in 2004 was 2970. The
corresponding cooling degree
days were 2000 and 2200 (base
10 ◦ C)
20 Puerto The cooling degree days (10 ◦ C)
Rico, USA are 5904, and the heating
degree days are negligible
21 Resolute, The heating degree days (18 ◦ C)
Canada are 12,571, and the cooling
degree days are negligible
Three temperature Calculation of the Residential building: the heating load for [35]
data sets: (a) WMO heating and cooling climatic scenario 3 decreased by 8–13% for
normal 1961–1990, (b) load of a residential climatic scenario C compared to A (+1 K
IEA design reference building and two mean annual temperature). The cooling
(1981–1990), and (c) offices. For all three load was negligible.
average Reference buildings, three Office buildings: the reduction of the
1984–2003) different scenarios of heating energy needs for climatic scenario
insulation are C compared to scenario A was 9–14% for
considered the first office building and 11–16% for
office 2. The relative increase of the cooling
load was low for both buildings
Climatic data for the Calculation of the For the 40 years, no increase trend for the [32]
period 1960–2000 temporal evolution of cooling degree days is observed. For the
the cooling degree days period 1981–2000 a small increasing
tendency
Use of climatic data for Calculation of the Small office: the cooling load between [36]
the period 1950–2010 annual heating and 1950 and 2010 increased by 8%, while the
cooling load for a heating load decreased by 30%. The total
small, medium and load increased by 3.5%
large building Medium office: the cooling load between
1950 and 2010 increased by 18%, while the
heating load is negligible.
Large office: the cooling load between
1950 and 2010 increased by 8%, while the
heating load decreased by 37%. The total
load decreased by 1.0%
Use of climatic data for Calculation of the total The total energy consumption for heating [37]
the period 1950–1990 energy consumption and cooling increased from 42.4 kWh/m2 /y
for heating and cooling to 47.7 kWh/m2 /y.
of a residential building
Data for the period Simulation of a small The variability of the total heating and [38]
1960–2004 office building cooling load was 9.2% for the whole period.
The maximum load was in 1990 and the
minimum in 2001. Not an important trend
was observed
Data for the period Simulation of a small The variability of the cooling load was 5% [38]
1960–2004 office building for the whole period. Not an important
trend was observed
Data for the period Simulation of a small The variability of the total heating load was [38]
1960–2004 office building 10% for the whole period

The variability of the heating and cooling load of a typical office
building during the period 1970–2010 in four major Greek cities
is investigated in [31]. Hourly data for Athens, Larisa, Heraclion
and Corfu are used to perform building energy simulations for
each individual year. A very important variation for both the heat-
ing and cooling loads was calculated as a result of the urban heat
island developed in the considered cities and the global climate
change. In particular, it is reported that the rate of temperature
increase at 14:00 during the summer period was 0.94 K/decade for
Athens, 0.5 K/decade for Larisa, 0.54 K/decade for Corfu and 0.42 K
for Heraclion. The corresponding temperature increase during the
winter period was 0.19 K/decade, 0.16 K/decade, 0.0 K/decade and
0.20 K/decade. The annual increase of the cooling degree hours dur-
ing this period was 97 for Athens, 51 for Larisa, 50 for Corfu and 39
for Heraclion while the corresponding annual decrease of the heat-
ing degree hours were 62, 98, 90 and 37. The significant warming
observed has resulted in a considerable increase of the cooling load
and a decrease of the heating load in all four cities. In particular,
the cooling load in Athens has increased by 28.6% or an increase
of 0.71 kWh/y, in Larisa it has increased by 15.5% or 0.33 kWh/y, in
Corfu, it has increased by 23.5% or an increase of 0.48 kWh/y and
finally in Heraclion, it has increased by 28.8% or 0.68 kWh/y. The
heating load in Athens, has decreased by 19.5% or 0.19 kWh/m2 /y,
in Larisa, it has decreased by 5.6% or 0.07 kWh/m2 /y, in Corfu, it
has decreased by 20% or 0.23 kWh/m2 /y, and finally in Heraclion, it
has decreased by 23% or 0.19 kWh/m2 /y. In all cases, the increase of
the cooling load was much higher than the corresponding decrease
of the heating load, thus, the urban warming has resulted to a sig-
nificant increase of the annual energy consumption of the office
buildings.
Climatic data for the period 1950–2010 for six major cities in
Cyprus are used in [33], to assess the possible increase of the cooling
load of a typical residential building. In most of the locations a very
significant increase of the ambient temperature as well as of the
cooling degree days were observed. In particular for Nicosia, it is
found that the temperature increase trend is of the order of 1.0 K
per 100 years. Also, it is calculated that the cooling degree days
(base 26 ◦ C), have increased from 610 in 1950 to 900 in 2010. The
corresponding increase of the cooling load was of the order of 33%.
In Limassol, the temperature increase trend was 1.3 K per 100 years,
and the cooling degree days increased from 600 in 1950 to 800 in
2010. The cooling load during the same period has increased by
 M. Santamouris / Energy and Buildings 82 (2014) 100–113
Fig. 3. Indices on the impact of heat island on the global energy consumption of a city.
43%. In Paphos, the cooling degree days increased from 250 in 1950
to 400 in 2010, while the corresponding increase of the cooling load
for the same period was close to 77%. In Larnaka, the cooling degree
days have increased from 630 to 870 in the period between 1950
and 2010, while the corresponding increase of the cooling load was
38%. In Famagusta, the cooling degree days increased from 620 to
820 between 1950 and 2010, while the cooling load increased by
33%. Finally, in Kerynia, cooling degree days increased from 400 in
1950 to 550 in 2010 and the cooling rise by 39 5%.
The impact of urban temperature increase on the energy con-
sumption of buildings for heating and cooling purposes for four
major cities in Switzerland, Zurich, Geneva, Lugano and Davos, is
investigated in [34]. Mean monthly temperature data covering the
period of 1901–2003 were used to calculate the heating and cooling
degree days for each city. The heating and cooling demands were
calculated as a function of the corresponding degree days for differ-
ent temperature bases. The gradient of the observed temperature
increase for the period 1901–2003, was 0.117 K/decade for Lugano,
0.143 K/decade for Zurich, 0.161 for Geneva and 0.165 K/decade
for Davos. The corresponding gradient increase for the period
1983–2003 was 0.795, 0.776, 0.795, and 0.642 K/decade. For the
same period 1983–2003, the increase of the heating degree days
(base 8.10 and 12 ◦ C), was 250–280 Hd/decade for Lugano, Davos
and Geneva and 250 per decade for Zurich, while the corre-
sponding increase of the cooling degree days (base 18.3 ◦ C), was
65.9 Cd/decade in Lugano 1.1 Cd/decade in Davos, 57.6 Cd/decade
in Geneva and 42.1 Cd/decade in Zurich. Energy calculations based
on the heating and cooling degree days, shown that for the period
1983–2003 the reduction of the heating load in Davos, was close
to 10% while the increase of the cooling load was negligible. In
Zurich, the heating load decreased by 13–16% and the cooling load
increased by 35%. In Geneva, the decrease of the heating load var-
ied between 15 and 21% and the increase of the cooling load was
close to 68%. Finally, for Lugano, the corresponding decrease of the
heating load was 18–25% and the increase of the cooling demand
was close to 45%.
111
An analysis of the possible increase of the heating and cooling
load of residential and office buildings because of the temperature
increase in Zurich, Switzerland is investigated in [35]. The existing
climatic analysis showed that for the period 1981–2003, the aver-
age annual temperature increase was 0.52 K/decade, while during
the summer and winter period it was slightly bit higher, 0.54 and
0.53 K/decade respectively. Three climatic data sets are used to
evaluate in a comparative way the energy impact of the observed
urban warming. The first climatic data set (A) was used as the ref-
erence one and included data for the period 1961–1990, the second
data set, (B) was for the period 1981–1990 and presented an aver-
age temperature increase of about 0.7 K compared to the first data
set. Finally, the third climatic data set, and (C) included data for the
period 1984–2003 while it presented an average annual tempera-
ture of 1 K higher than the climatic set (A). One residential and two
office buildings were considered and simulated. The second office
building presented higher internal gains than the first one. For all
three buildings, three different scenarios concerning the levels of
insulation, low, medium and high insulation levels, were consid-
ered. For the residential building, the obtained simulation results
showed that the heating load of the building as calculated using the
climatic scenario C, was 8–13% lower than that calculated using the
first climatic scenario. The overall simulation has shown that for the
low insulated building the heating days were 262 using scenario A
and reduced to 210 with scenario C. The corresponding values for
the medium insulation scenario were 222 and reduced to 189 and
for the highly insulated building 185 reduced to 154. The cooling
load of the residential building was negligible. For the first office
building, presenting the low internal gains, the calculated decrease
of the heating demand when scenario C was considered was 9–14%
compared to the loads calculated using the first climatic scenario.
The corresponding decrease of the heating load for the second office
building was 11–16%. For both office buildings the heating period
is shortened by 30–60 days per year. For all examined scenarios
and both buildings the cooling load was quite low and much below
10 kWh/m2 /y.
112 M. Santamouris / Energy and Buildings 82 (2014) 100–113
Fig. 4. Cooling energy consumption of all cases for 1970 against the corresponding consumption for 2010.
The possible increase of the energy consumption in Hong Kong
because of the urban climatic change is investigated in [32]. Tem-
perature data for the period 1960–2000 were analyzed and it is
found that when the data were grouped into four 10 year periods,
there was a 0.4 ◦ C rise in the dry bulb temperature from 1961–1970
to 1991–2000. The cooling degree days are calculated for different
temperature bases. Not a significant increase trend was observed
except for the years 1981–2000 where a very small increasing ten-
dency was found.
A detailed analysis of the variation of the heating and cooling
load of a small, medium and large building in Phoenix, Arizona, is
investigated for the period 1950–2010 [36]. Meteorological mea-
surements shown that the minimum temperature has risen about
0.47 ◦ C per decade, from 1960 to 2000. Detailed computer sim-
ulations are performed and it is found that for the small office
building the cooling load between 1950 and 2010 has increased
by 8%, while the heating load is decreased by 30% and the total load
is increased by 3.5%. For the medium office, the cooling load has
increased by 18%, while the heating load was negligible. Finally for
the large building, the cooling load increased by 8%, while the heat-
ing load decreased by 37%. The total energy load increased by 1.0%.
Another study aiming to evaluate the global impact of heat island in
Phoenix Arizona, is presented in [37]. The study compared the evo-
lution of the ambient temperature in an urban and a rural station.
For the urban station, the annual maximum temperature rate was
0.193 K/decade, while the annual minimum temperature rate is
0.984 K/decade. The values for the rural station were 0.03 K/decade
and 0.427 K/decade for the max and min temperatures respectively.
The study used data for the period 1950–1990 and calculated, using
dynamic simulation techniques, the total load of a residential build-
ing. It is found that the total energy consumption for heating and
cooling has increased from 42.4 kWh/m2 /y to 47.7 kWh/m2 /y.
An analysis of the evolution of the heating and cooling load for a
small office building for Washington CD, Puerto Rico and Resolute
Canada is performed in [38]. Climatic data for the period 1960–2000
are used. The analysis has shown that for Washington there is an
important increase of the cooling degree days along with a decrease
of the heating degree days. The heating and cooling load of a small
office building was calculated for each year using dynamic simula-
tion techniques. The temporal variability of the heating and cooling
load was 9.2% for Washington, 5% for Puerto Rico and 10% for Reso-
lute. The maximum values were calculated in 1990 in Washington,
1980 in Puerto Rico and 1972 in Resolute, while the corresponding
minimum values were presented at 2001 and 1976. Not an impor-
tant increasing or decreasing trend was observed.
Using the reported data on the specific temperature trends
and the corresponding energy demand, the cooling load of the
considered building is calculated for 1970 and 2010 using lin-
ear interpolation techniques for all the previously reported zones
except for Washington DC, Puerto Rico and Resolute, where the
warming trend was not defined. Fig. 4 gives the relation between
the cooling loads calculated using the climatic data of 1970 and
the corresponding values for 2010 for all cities where an important
cooling demand was calculated. As shown, there is a very signifi-
cant correlation between the two sets of values. As an average, and
on the basis of the available studies, it may be concluded that the
cooling load of the considered typical buildings has increased by
almost 23% during the period between 1970 and 2010. In paral-
lel, conclusions (d) and (e) of the chapter 2.1 are also valid for the
present analysis regarding the cooling demand of the buildings.
As it concerns the calculated reduction of the heating demand
of buildings caused by ambient warming, data from only six case
studies were available. Using exactly the same methodology as in
the case of the cooling load calculation, it was found that for the
six available case studies the average reduction of the heating load
caused by the urban warming for the period 1970–2010, is around
to 19%. Finally, the total heating and cooling load was calculated
for the six case studies and it is found that in overall, the energy
load for heating and cooling purposes during the period 1970–2010
increased by 11% because of the urban warming.
4. Conclusions
Urban heat island and global warming increase the temperature
of cities and exacerbate the energy demand of buildings. Energy
increase is very significant in cooling dominated zones where the
rise of the summer energy needs is much higher than the possible
decrease of the heating needs in winter. On the contrary in heat-
ing dominated zones, heat island may decrease the total annual
energy needs of buildings provided that buildings do not present
high cooling loads because of the significant internal loads.
Although the available studies comparing the energy consump-
tion of similar buildings located in urban and rural zones, are quite
limited, existing data reveals that the average increase of the cool-
ing load because of the heat island is statistically significant and in
average is close to 13%. In parallel, the annual cooling penalty per
 M. Santamouris / Energy and Buildings 82 (2014) 100–113
degree of average UHI intensity is very well correlated against the
logarithm of corresponding reference cooling demand.
Limited studies are also available regarding the global energy
penalty induced by the urban warming on the total building stock of
a city. Calculations for several cities around the world have shown
that the global energy impact of higher urban temperatures is quite
important. It is calculated that the average global energy penalty
per unit of city surface is significant (2.4 kWh/m2 ), while the aver-
age Global Energy penalty per unit of surface and degree of UHI
intensity is close to 0.74 kWh/m2 /K. In parallel, the average global
energy penalty per person is close to 237 kWh/p and the global
energy penalty per person and per degree of the UHI intensity is
around to 70 kWh/p/K. It is clear that these figures consist of pre-
liminary information may change when more data are available.
A significant increase of the energy consumption of buildings,
caused by the rise of the ambient temperature in cities, is calcu-
lated for the period 1970–2010. Based on a significant number
of studies it is calculated that the average increase of the cooling
demand of representative buildings during that period is close to
23%. In parallel, the corresponding average reduction of the heating
load is around to 19%, while during the same period, the average
total energy load of representative buildings spent for heating and
cooling purposes increased by 11%.
The specific studies and analysis make clear that urban warming
has a very significant impact on the global energy consumption of
buildings. The aim of the present investigation was to review the
existing knowledge and studies, evaluate and classify the existing
knowledge and calculate preliminary indicators and impact figures.
It is evident that the magnitude of the relevant estimations will
change every time a new study is available. However, the provided
information should be seen as a preliminary, yet accurate, analysis
describing existing knowledge and serving as the future basis for
further detailed studies.
Acknowledgment
I would like to express my gratitude to Prof. C. Cartalis for the
careful reading and the useful suggestions.
References
[1] M. Santamouris (Ed.), Energy and Climate in the Urban Built Environment,
James and James Science Publishers, London, 2001.
[2] M. Santamouris, Heat island research in Europe – the state of the art, Advances
in Building Energy Research 1 (2007) 123–150.
[3] M.J. Alcoforado, H. Andrade, Global warming and the heat island, in: J.M. Mar-
zluff, et al. (Eds.), Urban Ecology, Springer, Berlin, 2008.
[4] G. Mihalakakou, H.A. Flocas, M. Santamouris, C.G. Helmis, Application of neu-
ral networks to the simulation of the heat island over Athens, Greece, using
synoptic types as a predictor, Journal of Applied Meteorology 41 (5) (2002)
519–527.
[5] M. Santamouris, C. Cartalis, A. Synnefa, D. Kolokotsa, On the impact of urban
heat island and global warming on the power demand and electricity consump-
tion of buildings – a review, Energy and Buildings (2014) (in press).
[6] C. Cartalis, A. Synodinou, M. Proedrou, A. Tsangrassoulis, M. Santamouris, Mod-
ifications in energy demand in urban areas as a result of climate changes: an
assessment for the southeast Mediterranean region, Energy Conversion and
Management 42 (14) (2001) 1647–1656.
[7] E. Stathopoulou, G. Mihalakakou, M. Santamouris, H.S. Bagiorgas, Impact
of temperature on tropospheric ozone concentration levels in urban
environments, Journal of Earth System Science 117 (2008) 227–236
http://www.scopus.com/scopus/source/sourceInfo.url?sourceId=22666
[8] K. Pantavou, G. Theoharatos, A. Mavrakis, M. Santamouris, Evaluating thermal
comfort conditions and health responses during an extremely hot summer in
Athens, Building and Environment 2 (46) (2011) 339–344.
[9] K. Pantavou, G. Theoharatos, M. Santamouris, D. Asimakopoulos, Outdoor ther-
mal sensation of pedestrians in a Mediterranean climate and a comparison with
UTCI, Building and Environment 66 (2013) 82–95.
[10] M. Santamouris, D. Kolokotsa, On the impact of urban overheating and extreme
climatic conditions on housing energy comfort and environmental quality of
vulnerable population in Europe, Energy and Buildings (2014) (in press).
113
[11] A. Sakka, M. Santamouris, I. Livada, F. Nicol, M. Wilson, On the thermal per-
formance of low income housing during heat waves, Energy and Buildings 49
(2012) 69–77.
[12] M. Santamouris, K. Paraponiaris, G. Mihalakakou, Estimating the ecological
footprint of the heat island effect over Athens, Greece, Climate Change 80
(2007) 265–276.
[13] M. Santamouris, N. Papanikolaou, I. Livada, I. Koronakis, C. Georgakis, A.
Argiriou, D.N. Assimakopoulos, On the impact of urban climate to the energy
consumption of buildings, Solar Energy 70 (3) (2001) 201–216.
[14] S. Hassid, M. Santamouris, N. Papanikolaou, A. Linardi, N. Klitsikas, C. Georgakis,
D.N. Assimakopoulos, The effect of the Athens heat island on air conditioning
load, Energy and Buildings 32 (2) (2000) 131–141.
[15] D.K. Papanastasiou, D. Fidaros, T. Bartzanas, C. Kittas, Impact of urban heat
island development on buildings’ energy consumption, Fresenius Environmen-
tal Bulletin 22 (7A) (2013) 2087–2092.
[16] A. Fanchiotti, M. Zinzi, Impact of cool materials on urban heat islands and on
buildings comfort and energy consumption, in: Proc. ASES Conference, 2012.
[17] T.J. Chandler, The Climate of London, Hutchinson & Co Publishers Ltd., London,
UK, 1965.
[18] M. Kolokotroni, Y. Zhang, R. Watkins, The London heat island and building
cooling design, Solar Energy 81 (2007) 102–110.
[19] M. Kolokotroni, X. Ren, M. Davies, A. Mavrogianni, London’s urban heat island:
impact on current and future energy consumption in office buildings, Energy
and Buildings 47 (2012) 302–311.
[20] W. Briindl, P. Hoppe, Advantages and disadvantages of the urban heat island –
an evaluation according to the hygro-thermic effects, Archives for Meteorology,
Geophysics, and Bioclimatology, Series B 35 (1984) 55–66.
[21] M. Street, C. Reinhart, L. Norford, J. Ochsendorf, Urban heat island in Boston –
an evaluation of urban air-temperature models for predicting building energy
use, in: Proceedings of BS2013: 13th Conference of International Building Per-
formance Simulation Association, August 26–28, Chamb␫ry, France, 2013.
[22] F. Rong, Impact of Urban Sprawl on U.S. Residential Energy Use (Dissertation),
Faculty of the Graduate School of the University of Maryland, College Park,
2006 (In partial fulfillment of the requirements for the degree of Doctor of
Philosophy).
[23] D. Chen, Z. Ren, C.-H. Wang, M. Thatcher, X. Wang, Urban Heat Island on Aus-
tralian Housing Energy Consumption, Healthy Buildings, Brisbane, Queensland,
2012.
[24] Aecom: Economic Assessment of the Urban Heat Island Effect, Prepared by
AECOM Australia Pty Ltd., Level 9, 8 Exhibition Street, Melbourne VIC 3000,
Australia, 2012.
[25] H. Radhi, S. Sharples, Quantifying the domestic electricity consumption for air-
conditioning due to urban heat islands in hot arid regions, Applied Energy 112
(2013) 371–380.
[26] F. Skoulika, M. Santamouris, N. Boemi, D. Kolokotsa, On the thermal charac-
teristics and the mitigation potential of a medium size urban park in Athens,
Greece, Landscape and Urban Planning 123 (2014) 73–86.
[27] F. Chen, X. Yang, W. Zhu, WRF simulations of urban heat island under
hot-weather synoptic conditions: the case study of Hangzhou City, China,
Atmospheric Research 138 (2014) 364–377.
[28] Y. Hirano, T. Fujita, Evaluation of the impact of the urban heat island on residen-
tial and commercial energy consumption in Tokyo, Energy 37 (2012) 371–383.
[29] C. Li, J. Zhou, Y. Cao, J. Zhong, Y. Liu, C. Kang, Y. Tan, Interaction between urban
microclimate and electric air-conditioning energy consumption during high
temperature season, Applied Energy 117 (2014) 149–156.
[30] Z. Jiahua, Y. Fengmei, The characteristics of urban heat island variation in Beijing
urban area and its impact factors, in: Urban Remote Sensing Event, 2009 Joint,
2009, pp. 1–6, http://dx.doi.org/10.1109/URS.2009.5137502.
[31] J. Kapsomenakis, D. Kolokotsa, T. Nikolaou, M. Santamouris, S.C. Zerefos, Forty
years increase of the air ambient temperature in Greece: the impact on build-
ings, Energy Conversion and Management 74 (2013) 353–365.
[32] J.C. Lam, C.L. Tsang, D.H.W. Li, Long term ambient temperature analysis and
energy use implications in Hong Kong, Energy Conversion and Management
45 (2004) 315–327.
[33] M.A. Lange, P. Hadjinicolaou, M. Santamouris, G. Zittis, Assessment of increasing
electricity demand for space cooling under climate change condition in the
Eastern Mediterranean, in: 2nd Conference on Power Options for the Eastern
Mediterranean Region (POEM 2013), Nicosia, Cyprus, 2013.
[34] M. Christenson, H. Manz, D. Gyalistras, Climate warming impact on degree-days
and building energy demand in Switzerland, Energy Conversion and Manage-
ment 47 (2006) 671–686.
[35] T. Frank, Climate change impacts on building heating and cooling energy
demand in Switzerland, Energy and Buildings 37 (2005) 1175–1185.
[36] S. Doddaballapur, Analysis of the impact of urban heat island on building energy
consumption, in: Proc. 2012 ACES Solar Conference, World Renewable Energy
Forum (WREF), 2012.
[37] J.S. Golden, A. Brazel, J. Salmond, D. Laws, Energy and water sustainability the
role of urban climate change from metropolitan infrastructure, Journal of Green
Building 1 (3) (2006) 124–138.
[38] D.B. Crawley, Estimating the impacts of climate change and urbanization on
building performance, Journal of Building Performance Simulation 1 (2) (2008)
91–115.