Solar Energy 81 (2007) 358–368

www.elsevier.com/locate/solener

Daytime urban heat islands from Landsat ETM+ and Corine land
cover data: An application to major cities in Greece
Marina Stathopoulou *, Constantinos Cartalis
Remote Sensing and Image Processing Laboratory, Division of Applied Physics, Department of Physics, Building PHYS-5,
University Campus, 157 84 Athens, Greece

Received 4 October 2005; received in revised form 16 June 2006; accepted 21 June 2006
Available online 28 August 2006

Communicated by: Associate Editor Matheos Santamouris

Abstract

Satellite images in the thermal infrared can be used for assessing the thermal urban environment as well as for defining heat islands in
urban areas. In this study, the thermal environment of major cities in Greece (Athens, Thessaloniki, Patra, Volos and Heraklion) is
examined using satellite images provided by the Landsat Enhanced Thematic Mapper (ETM+) sensor on board Landsat 7 satellite cor-
responding to the daytime and warm period when the surface urban heat island (SUHI) phenomenon is best observed. The spatial struc-
ture of the thermal urban environment is analyzed in each case study and the ‘‘hottest’’ surfaces within the urban settings are identified
and related to the urban surface characteristics and land use. For the needs of the study, the Corine land cover (CLC) database for
Greece is also used, in an effort to define more effectively the link between surface emissivities, land surface temperatures and urban sur-
face characteristics.
 2006 Elsevier Ltd. All rights reserved.

Keywords: Remote sensing; Urban heat island; ETM+; Land cover; CORINE

1. Introduction the layer closest to the surface in cities consisting of the

air contained between the urban roughness elements
Urban heat island (UHI) phenomenon describes the (mainly buildings) and the urban boundary layer (UBL),
excess warmth of the urban atmosphere and surfaces com- which is the layer of the atmosphere situated directly above
pared to the non-urbanized rural surroundings. In general, the urban canopy layer. These two distinct layers and heat
three types of heat islands are recognized: islands are governed by different processes; UBL is domi-
nated by processes acting at the local or meso-scale, whereas
• the canopy layer heat island UCL is a micro-scale concept and its climate is affected by
• the boundary layer heat island, and the nature of the immediate site. Finally, the surface urban
• the surface urban heat island. heat island refers to the relative warmth of the urban sur-
faces compared to their non-urbanized surroundings.
The first two types are atmospheric heat islands pro- Heat islands development can be attributed to a number
duced by urbanization. Both refer to a warming of the of factors including construction materials of high heat
urban atmosphere, which is divided into two layers accord- capacity and low solar reflectivity such as asphalt and con-
ing to Oke (1976): the urban canopy layer (UCL), which is crete, reduced turbulent heat transfer and long-wave radia-
tive heat loss due to street canyons geometry, reduced
*
Corresponding author. Tel.: +30 1 727 6843; fax: +30 1 727 6774. latent heat loss by evaporation due to the replacement of
E-mail address: mstathop@phys.uoa.gr (M. Stathopoulou). the natural green surfaces with dry surfaces and increased

0038-092X/$ - see front matter  2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.solener.2006.06.014
 M. Stathopoulou, C. Cartalis / Solar Energy 81 (2007) 358–368 359

Nomenclature

BT at-sensor brightness temperature (K) ETM+ Enhanced Thematic Mapper Plus: the sensor
Corine COoRdinate INformation on the Environment onboard Landsat 7 satellite
CLC Corine land cover e surface emissivity
DN digital number values K1 calibration constant (666.09 Wm2 sr1 lm1)
DT intensity of the surface urban heat island (C) K2 calibration constant (1282.71 K)
DTur surface temperature difference between urban L spectral radiance (Wm2 sr1 lm1)
and rural areas NDVI normalized difference vegetation index
DTsr surface temperature difference between subur- SUHI surface urban heat island
ban and rural areas Ts land surface temperature (C)
DTmr surface temperature difference between mixed
urban and rural areas

anthropogenic heat emissions into the urban atmosphere. land cover (CLC) database in a GIS framework. CLC was
The intensity of the UHI (DT) is reflected by the tempera- finalized in the early 1990s as part of the European Com-
ture differences between urban and rural sites and depends mission programme to COoRdinate INformation on the
on the size, population and industrial development of a Environment (Corine) and it provides consistent, reliable
city, topography, physical layout, regional climate and and comparable information on land cover across Europe.
meteorological conditions (Oke, 1987). The novelty of the methodology presented lies in consider-
Voogt and Oke (2003) review the use of thermal remote ing spatially-averaged surface temperatures per land cover
sensing for the study of urban climates with respect to the compared to point surface temperature data for inclusion
heat islands and describe the distinction between the atmo- in the measurement of the surface UHI intensity between
spheric and the surface UHIs. Atmospheric UHIs are usu- the urban land covers and rural surrounds. In addition,
ally detected by ground-based air temperature the incorporation of the Corine land cover database in the
measurements taken from standard meteorological sta- surface UHI measurement makes the method applicable
tions, whereas surface UHIs are observed from thermal to all European cities included in the CLC database and
remote sensors which record the upwelling thermal radi- minimizes the number of assumptions considered or even
ance emitted by the surface area that lies within the instan- the computational time needed from processing satellite
taneous field of view (IFOV) of the sensor. In contrast to data in case there is no information about the land cover
atmospheric UHIs that are best expressed under calm distribution of a studied area.
and clear conditions at night, surface UHIs are usually The methodology is applied to Athens (3758 0 N,
studied by using satellite or aircraft thermal data of high 2346 0 E), the capital city of Greece, as well as to the cities
spatial resolution (6100 m) acquired at daytime when heat of Thessaloniki (4038 0 N, 2258 0 E), Volos (3924 0 N,
island intensities are greatest (Roth et al., 1989). 2259 0 E), Patra (3814 0 N, 2147 0 E) and Heraklion
Satellite thermal data can effectively depict the patterns (3520 0 N, 2512 0 E), which are the four largest cities of
of the thermal environment of extensive urban areas on a Greece, in order to analyze the spatial structure of their day-
repeated basis. Thus, spatial coverage and temporal repeti- time thermal environment and estimate their surface UHI
tion are the main advantages of using satellite thermal intensities. All of them, termed as Olympic Cities during
remote sensing technique in the study of the urban cli- the Olympic Games of 2004, are regional capitals and have
mates. Many surface UHI studies have been conducted large ports of intense trade. Previous studies on their urban
using thermal data from satellites (Aniello et al., 1995; environment using satellite (Stathopoulou et al., 2004) or
Streutker, 2002; Lo and Quattrochi, 2003; Dousset and ground data (Santamouris et al., 2001; Livada et al., 2002;
Gourmelon, 2003). These studies, employing medium scale Mihalakakou et al., 2004) have shown that the urban heat
sensors (Landsat TM, NOAA AVHRR), give a spatially island of these cities is pronounced and of comparative
continuous view of the surface UHI over large urban areas intensity; night-time UHI intensity can reach up to 8 C
than is feasible using data from meteorological station net- for the cities of Thessaloniki and Heraklion and up to
works and recognize the correlation between land surface 7 C for the cities of Patra and Volos, whereas the night-time
temperature and land cover using land cover maps of the UHI intensity for the central Athens area can rise up to 5 C.
study area extracted from satellite data (SPOT HRV, Fig. 1 illustrates the location of the five Greek harbor cities.
Landsat TM, Landsat MSS).
This study presents a methodology for estimating the 2. Methodology
surface UHI intensity of urban areas using Landsat 7 ther-
mal images of high spatial resolution (60 m) in combination The methodology for mapping and estimating surface
with urban land cover information provided by the Corine UHI intensities, as shown in Fig. 2, involves:
360 M. Stathopoulou, C. Cartalis / Solar Energy 81 (2007) 358–368

Fig. 1. Map of Greece showing the location of the cities under study.

Fig. 2. Flowchart of the method leading to the estimation of the surface UHI intensity from satellite and Corine data.
 M. Stathopoulou, C. Cartalis / Solar Energy 81 (2007) 358–368 361

• Landsat ETM+ processing for obtaining at-sensor Table 1

brightness temperatures (BT), Landsat ETM+ images used in this study
• CLC data processing for the spatial definition of City Date (d/m/y) Time (UTC)
the urban, suburban, mixed and rural regions of the Athens 20/05/2000 08:57
city, which allows further the more spatially accu- Thessaloniki 30/05/2001 09:00
rate determination of their respective surface emissivi- Volos 24/08/2000 08:56
Patra 28/06/2000 09:02
ties (e), Heraklion 09/07/2000 08:45
• Land surface temperature (Ts) mapping,
• Surface UHI intensity (DT) estimation based on the spa-
tially-averaged surface temperature differences between over the region of Greece is near solar noon time
the different land cover types of the city (urban/densely (12:00 am).
built, suburban/medium built, mixed urban area, rural Thermal band image data calibration is performed in a
area). two-step process as proposed by the Landsat Project Sci-
ence Office (2002): (a) conversion of the digital number
(DN) values of band 6 into spectral radiance L
2.1. Landsat ETM+ processing (Wm2 sr1 lm1) using the following equation:
L ¼ 0:0370588  DN þ 3:2 ð1Þ
Landsat 7 satellite has a near polar sun-synchronous
orbit, meaning that it orbits the earth from the north to and then (b) conversion of the spectral radiance L to at-
south poles and vice versa and covers each place on Earth sensor brightness temperature BT in Kelvin. The conver-
at a constant local time of day. The revisit time of the sion formula is given by:
satellite is 16 days. The ETM+ sensor on board the satel- K2
lite can provide image data from the visible to thermal BT ¼  K 1  ð2Þ
‘n L þ 1
infrared spectral regions (Fig. 3). Landsat 7 was chosen
in this study, because it collects thermal measurements where BT is the at-sensor brightness temperature in K; K2
with the highest spatial resolution currently available from is the calibration constant (1282.71 K); K1 is the calibration
space. constant (666.09 Wm2 sr1 lm1) and L is the spectral
Landsat 7 ETM+ daytime thermal images during radiance at-sensor (Wm2 sr1 lm1).
the warm period are used in order to map the thermal
urban environment of the cities of Athens, Thessaloniki, 2.2. Corine land cover data processing
Patra, Volos and Heraklion. ETM+ thermal band 6
(10.4–12.5 lm) has a spatial resolution of 60 m at nadir, Corine land cover (CLC) is a geographic land cover/
which is considered to be suitable for capturing the com- land use database for a pan-European region. It is pro-
plex intra-urban surface temperature differences allowing duced by the European Environment Agency and provides
thus, an effective and detailed analysis of the urban cli- consistent information on land cover across Europe using a
mate. Table 1 lists the acquisition date and time of the nomenclature of 44 standard classes which are organized in
Landsat ETM+ images recorded over the selected cities. 3 levels of detail, as shown in Fig. 4 (European Commis-
All images correspond to highly clear atmospheric condi- sion, 1993). The CLC90 database for Greece in vector for-
tions and warm season. They were acquired through the mat is utilized in this study in an effort to link more
US Geological Survey (USGS) Earth Resource Observa- effectively land cover and surface emissivity.
tion Systems Data Center and are the best images avail- For this purpose, the original 44 classes of Corine
able, given that the local time of the satellite overpasses were grouped into only 5 ones: ‘‘urban/densely built’’,

Fig. 3. Landsat 7 orbit (left) and ETM+ sensor characteristics (right). Bandwidths are given in microns (lm), whereas spatial resolution in meters (m).
362 M. Stathopoulou, C. Cartalis / Solar Energy 81 (2007) 358–368

Fig. 4. Original Corine and grouped land cover classes.

‘‘suburban/medium built’’, ‘‘mixed urban area’’, ‘‘rural In the following, emissivity correction according to land
area’’ and ‘‘water surface’’; thus, a new land cover polygon cover is performed in order to compute emissivity cor-
map for each city was produced (Fig. 5). As our interest rected land surface temperatures as described next. Ignor-
focuses mainly on urban areas, the Agricultural areas and ing emissivity differences between land covers contributes
Forests and semi-natural areas of the level 1 CLC legend to error in the derivation of actual surface temperatures
are united to form the ‘‘rural area’’ class. Similarly, Wet- and to subsequent inaccuracies when surface temperature
lands and Water bodies were grouped together and classi- variations and relative comparisons are examined as is
fied as ‘‘water surface’’. However, the 11 land cover/use the case with urban heat island studies. An error in surface
classes included in the Artificial surfaces of the CLC legend emissivity of 0.01 causes an error in surface temperature of
are re-organized and grouped into new classes based on about 2 K for conditions that are typical of the earth’s sur-
similar land cover structure characteristics. The grouping face (i.e., 260 K 6 Ts 6 320 K; 0.9 6 e 6 1.0) (Prata,
performed, as shown in Fig. 4, is aiming at: (a) linking 1993).
more effectively land cover and surface emissivity which
allows for emissivity corrected surface temperatures to be 2.3. Land surface temperature retrieval
computed, thus reducing the error produced in measuring
surface temperature differences between land cover types Brightness temperature derived from Eq. (2) is the
(b) helping to spatially distinguish between the residential temperature that a blackbody would obtain in order to
areas and the urban use areas of the city and (c) defining produce the same radiance at the same wavelength
classes related to the urban heat island effect. In this con- (k = 11.5 lm). Therefore, additional correction for spectral
text, Discontinuous urban fabric and Green urban areas of emissivity is required to account for the non-uniform emis-
the CLC legend are grouped together classified as ‘‘subur- sivity of the land surface. Emissivity correction is carried
ban/medium built’’. Likewise, Industrial/commercial and out using surface emissivities for the specified land covers
transport units, Mine, dump and construction sites as well (Table 2) derived from the methodology described in
as Sport and leisure facilities of the CLC legend are united Stathopoulou et al. (2006). The procedure involved com-
into one class named as ‘‘mixed urban area’’. bining surface emissivity maps obtained from the Normal-
 M. Stathopoulou, C. Cartalis / Solar Energy 81 (2007) 358–368 363

Fig. 5. Land cover maps of: (a) Athens; (b) Thessaloniki; (c) Patra; (d) Heraklion and (e) Volos as reproduced from the Corine land cover data.

Table 2 emissivity corrected land surface temperature Ts is com-

Surface emissivity values by land cover type puted as follows:
Land cover type Emissivity
BT
Urban/densely built 0.946 Ts ¼ n h io ð3Þ
kBT
Suburban/medium built 0.964 1þ q
 ‘ne
Mixed urban area 0.950
Rural area 0.980 where Ts is the land surface temperature (K); BT is at-sensor
Water surface 0.990
brightness temperature (K); k is the wavelength of emitted
radiance (11.5 lm); q is (h · c/r) = 1.438 · 102 m K; e is
the spectral surface emissivity.
ized Difference Vegetation Index Thresholds Method Accurate determination of the surface temperature also
(NDVITHM) (Sobrino and Raissouni, 2000) with land requires correction for atmospheric interference. How-
cover information from the Corine database. Thus, the ever, the error in surface temperature due to an error in
364 M. Stathopoulou, C. Cartalis / Solar Energy 81 (2007) 358–368

the emissivity correction is two times larger than the one due mean surface temperatures with standard deviations are
to an error in the atmospheric correction (Prata et al., 1995). computed for each land cover class by using a polygon
Taken that this study is interested in relative surface temper- summary Geographical Information Systems (GIS) opera-
ature differences between urban and rural regions, the error tion. In this way, the intensity of the surface UHI (DT) is
due to the atmospheric effects is not corrected. Since the estimated from the surface temperature differences between
UHI effect is studied in a spatial extent of a few kilometers the different urban regions (as described by the different
around the city center (local scale), it is reasonable to assume urban land covers) and the surrounding rural area of the city.
the homogeneity of the atmosphere over the urban area. The
error produced in the land surface temperature subject to the 3. Application on Greek cities
influence of the atmospheric conditions is relatively uniform
across the image and it is adjusted in case that surface tem- The methodology presented above is applied to the
perature differences are examined. metropolitan city of Athens (745,514 inhabitants) and to
the major cities of Greece: Thessaloniki (363,987 inhabit-
2.4. Estimation of the surface UHI intensity ants), Patra (160,400 inhabitants), Volos (82,439 inhabit-
ants) and Heraklion (130,914 inhabitants) in order to
The surface UHI intensity estimation is performed by examine their daytime thermal environment during the
combining the surface temperature data (in raster format) warm season as well as to detect areas of intense thermal
with the reproduced land cover data (in vector format). In radiation (hot spots) within the urban settings. Further-
particular, the land cover map (polygon layer) and the sur- more, the daytime surface UHI intensity of each city is
face temperature image of the city are superimposed and estimated.

Fig. 6. Land surface temperature map of Athens (32 · 20 km).

 M. Stathopoulou, C. Cartalis / Solar Energy 81 (2007) 358–368 365

Fig. 7. Land surface temperature map of Thessaloniki (20 · 16 km).

Figs. 6–10 illustrate the spatial distribution of Ts values 2. Solar heating of the open spaces and bare soil associated
for the selected cities. Dark to bright tones indicate cooler with mines, dump sites as well as with construction sites
to warmer land surface temperatures. Mean surface tem- for the Olympic Games of 2004.
peratures by land cover class are given in Table 3, whereas
surface UHI intensities of the selected cities are presented Visual inspection of the spatial distribution of Ts for the
in Table 4, where DT denotes the surface temperature Athens basin (Fig. 6) indicate that southwestern suburbs
difference in between urban–rural (u–r), suburban–rural such as Petroupoli, Nikaia, Kallithea etc. obtain higher
(s–r) and mixed-rural (m–r) areas. surface temperatures than northeastern suburbs such as
Kifissia, Chalandri, Zografos etc. This can be explained
4. Results and discussion by the fact that in these suburbs the vegetation cover and
density is very low. In addition, the industrial area located
It is clear from Table 3 that mixed urban areas exhibit in between Nikaia, Kallithea and the centre of Athens
high surface temperature values for all cities. Results appears to exhibit high surface temperatures. Most of the
of the estimated daytime UHI intensities for the city of industrial activities are gathered to this area consisting of
Athens on the 20th of May, 2000 show that central urban textile, cement, chemical, fertilizer, paint and paper facto-
areas are 3.3 C warmer than rural areas, whereas subur- ries. Except from industrial plants, the area has many bus
ban areas are about 2.3 C warmer than rural areas. It is and train depots as well as large tracts of derelict land.
also found that mixed urban areas are ‘‘hot spots’’ of the The high surface temperatures observed at this industrial
city as they appear to be 1.9 C warmer than the central zone can be attributed to the local structure development
urban areas and 5.2 C warmer than surrounding rural comprising of low-rise buildings along wide paved streets
areas. This can be attributed to: and extensive unshaded open spaces consisting of non-
build areas and bare ground. Finally, high surface temper-
1. The increased proportion of the surfaces covered with atures are also detected at the south-east edge of the basin,
asphalt and concrete in industrial/commercial and where the Elliniko Airport is located due to the solar heat-
transport units as well as to the surface/building materi- ing of the extended paved surfaces associated with the air-
als used in these areas. port facilities. Therefore, the industrial zone of Athens as
366 M. Stathopoulou, C. Cartalis / Solar Energy 81 (2007) 358–368

Fig. 8. Land surface temperature map of Patra (16 · 13 km).

Fig. 9. Land surface temperature map of Volos (12 · 16 km).

well as the Elliniko Airport can be clearly identified as ‘‘hot highest surface temperatures of the order of 34 C. How-
spots’’ of the city. ever, they present a weak surface temperature difference
In the case of Thessaloniki, an UHI intensity of the from the urban/densely built areas of about 0.5 C, which
order of 2.7 C is estimated between the central urban can be attributed to the built form of the city. Thessaloniki
and surrounding rural areas of the city on the study day is a major harbor city and its urban structure has been
(30/5/2001). Suburban areas appear to be 1.2 C cooler developed in such a way so as to subserve the function of
than the central urban areas and 1.5 C warmer than the the port. For this reason, the main road network and com-
surrounding rural areas. Mixed urban areas obtain the mercial centre are located along the coastal zone of the city
 M. Stathopoulou, C. Cartalis / Solar Energy 81 (2007) 358–368 367

Fig. 10. Land surface temperature map of Heraklion (10 · 15 km).

Table 3 30.7 C for the surrounding rural areas to 32.2 C for the
Mean values of Ts (in C) by land cover type suburban/medium built areas and up to 33.4 C for the
City Urban/ Suburban/ Mixed Rural Water urban/densely built areas of Patra. Therefore, a daytime
densely medium urban area surface UHI intensity of the order of 3 C is estimated between
built built area
urban and rural regions, whereas suburban/medium built
Athens 33.5 (1.5) 32.5 (1.4) 35.4 (2.0) 30.2 (3.1) 18.3 (1.9) areas are 1.7 C cooler than central urban areas. Also in this
Thessaloniki 33.4 (1.9) 32.2 (1.1) 34.0 (1.4) 30.7 (2.1) 19.9 (1.5)
Patra 33.8 (3.3) 32.1 (2.5) 35.4 (1.9) 30.8 (2.8) 20.7 (0.6)
case, it is evident the influence of the port area to the ther-
Volos 36.5 (1.4) 35.3 (1.7) 38.4 (2.2) 36.1 (3.5) 24.4 (1.7) mal environment of the city. In addition, the industrial area
Heraklion 39.9 (4.9) 38.3 (4.3) 45.3 (3.9) 38.0 (2.0) 25.2 (0.9) located northeastern of Patra, between the suburbs of Rio
Standard deviation values are also given in parenthesis. and Agios Vasilios, obtains high surface temperature values
of about 35.4 C and appears to be 1.6 C and even 4.6 C
warmer than central urban and surrounding rural areas.
Table 4 In the city of Volos, estimated daytime UHI intensities
Estimated surface UHI intensities DT (in C) are weaker; urbanized areas are 0.4–0.8 C warmer than
City DTur DTsr DTmr rural areas, a fact which can be attributed to the solar heat-
Athens 3.3 2.3 5.2 ing of the low agriculture in the vicinity of Volos. Lower
Thessaloniki 2.7 1.5 3.3 surface temperatures are observed in the northeastern parts
Patra 3.0 1.3 4.6 of the city where suburbs are located on the forested slope
Volos 0.4 0.8 2.3
Heraklion 1.9 0.3 7.3
of Mount Pilion e.g. Portaria (Fig. 9). On the other hand,
mixed urban areas that include industrial units are ‘‘hot
spots’’ of the city and appear to be 2.3 C warmer than
that extends from the Eleutherio-Kordelio and the port the surrounding rural area.
area to Kalamaria at the north-eastern parts of the city Finally, the highest surface temperatures are observed
(Fig. 7). As a result, the urban fabric extents from the over the city of Heraklion. Results of the estimated day-
coast, where most of the high-rise and high-density residen- time UHI intensities for the 9th of July, 2000 show that
tial buildings exist, to the inland of higher elevation, where urban/densely built areas, located in the vicinity of the port
urban density is lower and vegetation cover is higher. As area, obtain high surface temperatures of about 40 C, that
shown in Fig. 7, the thermal pattern of Thessaloniki is, 1.9 C warmer than surrounding rural areas. Suburban/
appears to follow the physical layout of the city. Warmer medium built areas are 1.6 C cooler than urban/densely
areas are detected near the coast and they are related to built areas presenting a mean surface temperature value
urban/densely built and mixed urban areas, whereas cooler of 38.3 C. From the surface temperature distribution
areas are correlated with the more green suburbs of Stau- map of Heraklion (Fig. 10), it is evident that maximum sur-
roupolis, Polikhni, Triandria and Pilaia, in the peripheral face temperatures are detected in Nea Alikarnassos, where
zone of the city. the airport of Heraklion is located. In fact, the airport area
Likewise, in the city of Patra a surface temperature appears to be 5.4–7 C warmer than urbanized areas of
increase is observed in the transaction from the rural to Heraklion and 7.3 C warmer than surrounding rural
the urban regions where maximum surface temperatures areas, therefore, it can be clearly identified as a ‘‘hot spot’’
are detected (Fig. 8). Surface temperature values range from of the city.
368 M. Stathopoulou, C. Cartalis / Solar Energy 81 (2007) 358–368

It must be mentioned that this study was performed by References

using Landsat ETM+ data on individual dates of the warm
season for each city. Considering that the UHI effect exhib- Aniello, C., Morgan, K., Busbey, A., Newland, L., 1995. Mapping micro-
its great spatial and temporal fluctuations and that it urban heat islands using Landsat TM and a GIS. Computers and
Geosciences 21 (8), 965–969.
strongly depends on weather conditions and synoptic flow Dousset, B., Gourmelon, F., 2003. Satellite multi-sensor data analysis of
patterns, future studies may be focused on the application urban surface temperatures and landcover. Isprs Journal of Photo-
of this method to a time series of thermal daytime satellite grammetry and Remote Sensing 58 (1–2), 43–54.
data during the warm season (May–September) as well as European Commission, 1993. CORINE land cover; guide technique,
to the development of a European consistent method for Report EUR 12585EN. Office for Official Publications of the Euro-
pean Communities, Luxembourg.
the assessment of the surface UHI effect. Landsat Project Science Office, 2002. Landsat 7 science data user’s
handbook. Goddard Space Flight Center, Available from: www
5. Conclusions address: http://ltpwww.gsfc.nasa.gov/IAS/handbook/handbook_toc.
html.
Livada, I., Santamouris, M., Niachou, K., Papanikolaou, N., Mihalaka-
In this study, a methodology for estimating the daytime
kou, G., 2002. Determination of places in the great Athens area where
surface UHI intensity of urban areas was presented. The the heat island is observed. Theoretical And Applied Climatology 71,
method employed Landsat 7 daytime thermal images of 219–230.
high spatial resolution (60 m) and urban land cover infor- Lo, C.P., Quattrochi, D.A., 2003. Land-use and land-cover change, urban
mation provided by the Corine land cover (CLC) database, heat island phenomenon, and health implications: a remote sensing
which were combined in a GIS framework. In conclusion, approach. Photogrammetric Engineering and Remote Sensing 69 (9),
1053–1063.
the application of the method has been proven successful in Mihalakakou, G., Santamouris, M., Papanikolaou, N., Cartalis, C.,
terms of the use of ETM+ images (in the thermal infrared) Tsangrassoulis, A., 2004. Simulation of the urban heat island
for the identification and the spatial analysis of the daytime phenomenon in Mediterranean climates. Pure and Applied Geophysics
UHI effect. Taken the adequate spatial resolution of the 161, 429–451.
images (60 m), the method may be also used for: Oke, T.R., 1976. The distinction between canopy and boundary-layer
urban heat islands. Atmosphere 14, 268–277.
Oke, T.R., 1987. Boundary Layer Climates. Methuen, London, New
• the depiction of ‘‘hot spots’’ which may have been devel- York.
oped, developed, Prata, A.J., 1993. Land surface temperatures derived from the advanced
• the monitoring of changes in urban climate, as well as very high resolution radiometer and the along-Track Scanning
Radiometer 1. Theory. Journal of Geophysical Research 98 (D9),
• for recording the thermal behavior of urban surfaces in
16,689–16,702.
relation to surface characteristics. Prata, A.J., Caselles, V., Coll, C., Sobrino, J., Ottle, C., 1995. Thermal
remote sensing of land surface temperature from satellites: current
Some of the advantages of the method proposed status and future prospects. Remote Sensing Reviews 12, 175–224.
include: Roth, M., Oke, T.R., Emery, W.J., 1989. Satellite-derived urban heat
islands from 3 coastal cities and the utilization of such data in urban
climatology. International Journal of Remote Sensing 10 (11), 1699–
• wide applicability, 1720.
• broad urban coverage, Santamouris, M., Papanicolaou, N., Livada, I., Koronakis, I., Georgakis,
• regular revisit capabilities, resulting in up-to-date C., Argiriou, A., Assimakopoulos, D.N., 2001. On the impact of urban
information, climate on the energy consumption of buildings. Solar Energy 70 (3),
• ability to combine satellite thermal data with other dig- 201–216.
Sobrino, J.A., Raissouni, N., 2000. Towards remote sensing methods for
ital data (such as demographic data, topography, popu- land cover dynamics monitoring. Application to Morocco. Interna-
lation density data, electric power consumption data, tional Journal of Remote Sensing 21, 353–366.
etc.), Stathopoulou, M., Cartalis, C., Keramitsoglou, I., 2004. Mapping micro-
• use of cost effective data. urban heat islands using NOAA/AVHRR images and CORINE land
cover: an application to coastal cities of Greece. International Journal
of Remote Sensing 25 (12), 2301–2316.
On the other hand, a disadvantage of the method may Stathopoulou, M., Cartalis, C. Petrakis, M., 2006. Integrating CORINE
be considered the sun-synchronous orbit characteristic, a land cover data and Landsat TM for surface emissivity definition: an
fact which limits the applicability of the method at a time application for the urban area of Athens, Greece, International
other than the overpass time of the Landsat 7 satellite. Journal of Remote Sensing, in press.
However, despite this limitation, it is clear that this method Streutker, D.R., 2002. A remote sensing study of the urban heat island of
Houston, Texas. International Journal of Remote Sensing 23 (13),
may produce results that can be effectively combined with 2595–2608.
conventional field surveys for a complete assessment of Voogt, J.A., Oke, T.R., 2003. Thermal remote sensing of urban climates.
the urban heat island effect. Remote Sensing of Environment 86 (3), 370–384.