The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B7. Beijing 2008 
channels of the Landsat satellite sensor images were used for 
1987, 1997 and 2007 years. Density slice algorithm was used 
for classification of the image which is Landsat satellite’s 
thermal channel. Little ranges were used for digital number. 
Moreover, classification of the urban, maximum likelihood 
method which is a supervised control system was used. More 
than 30 classes were chosen at the beginning of the study for the 
classification of the urbanization. Then, the number of the 
classes was decreased to 5 basic classes and these basic classes 
have been shown in different colors. In addition, meteorological 
data which are temperature, precipitation, humidity, for the 
same period which are 16 July 1987, 18 July 1997 and 28 June 
2007, were used. The images that we have are classified for 
thermal channel; also the temperature knowledge and the 
coordinate knowledge of the stations have been added to this 
classification. Urban heat islands were created by the 
meteorological data and coordinate data for Istanbul. Different 
major centers for urban heat island are found in Istanbul. Our 
major finding is that expansion of urban areas in Istanbul lead 
growth in thermal radiation of land surface in highly dense areas. 
This is mainly due to unplanned and unorganized rapid 
urbanization. 
2. STUDY AREA AND DATA 
2.1 Study Area 
The city of Istanbul, which is located at 41,01 °N, 28,58 °E in 
Asia and Europe, is the largest city of Turkey with a population 
of over 12,5 million (DIE, 2008). Istanbul’s population was 
nearly doubled in 20 years between 1980 and 2000, the fastest 
growth period. For the period between 1990 and 2000, the 
population growth rate of Istanbul is 29,64% for urban parts and 
81% for rural parts of the city. Total population growth rate is 
33.1% for the same period. In near future its population is 
expected to reach 20 million by 2030 (Turkstat, 2008). The 
Bosphorus, a 30-km strait that connects the Black Sea with the 
Sea of Marmara, is considered to be the boundary between 
Europe and Asia, and the urban Istanbul is located on both sides 
of the southern half of the strait. The north of the city towards 
the Black Sea is mostly covered by protected forest patches, and 
the expansion of the city in that direction is mostly confined to 
along the Bosphorus. The most densely populated parts of the 
city lie in the south along the Sea of Marmara. 
UHI in Istanbul has been previously studied by several 
researchers using long-term temperature data from 
meteorological stations within and around the city. In this work, 
we use the thermal remote sensing to study of urban climate in 
Istanbul from 1987 to 2007 for the first time. Thermal and other 
channels of the Landsat satellite sensor images were used for 
1987, 1997 and 2007 years. 
2007 Landsat 5 TM 
Figure 1. Study area, Istanbul 2007 (RGB : 321) 
2.2 Landsat Data 
Landsat Thematic Mapper satellite’s images have the 120m 
spatial resolution of thermal infrared band, and the other bands 
have 30m spatial resolution. The dates of the images are; 
25 September 1987 
18 July 1997 
28 June 2007 
2.3 Meteorological Data 
Temperature data which are belongs to 28 June, 2007, provided 
by meteorological stations in Istanbul, are used. There are 20 
meteorological stations in Istanbul, some of them are belong to 
Turkish state meteorological service and some of them are 
automatic weather observation stations (AWOS) which are 
belong to the greater Istanbul Municipality. The automatic 
weather observation stations are started up in 2006 in Istanbul. 
The stations are Gatalca, Kumkoy, Sanyer-Kire9bumu, Goztepe, 
§ile, Bahfekoy, Florya, Kartal, Kandilli, Istanbul University, 
§ile, Kartal, Gebze, Samandira stations, Atatiirk and Sabiha 
Gok<?en Airport stations, £avu§ba§i, Omerli, Btiyiikada, 
Olimpiyat, Aksaray, Terkos, AKOM, £anta, Kamiloba, 
Hadimkoy stations. 
3. METHODOLOGY 
3.1 Classification 
3.1.1 Urban Classification 
Firstly, 1, 2, 3, 4, 5 and the 7. bands of the Landsat 5 TM 
satellite are classified to show the urbanization in Istanbul for 
last 20 years. Maximum Likelihood classification method, 
which is a supervised method, is used. More than 30 classes 
were chosen at the beginning of the study for the classification 
of the urbanization. Then, the number of classes was decreased 
to 7 basic classes which are urban, road, agriculture, sea, lakes, 
forest, sand, and the eighth is cloud. These basic classes have 
been shown in different colors. 
First image of the classification is shown in figure 2, and it 
belongs to 25 September, 1987. This classification has 8 classes 
which are urban area, agricultural area, forest, sea, lake, sand 
area, roads and clouds. Figure 3 belongs to 18 July, 1997 
image’s classification. There is no cloud in this classification, 
and it has 7 classes which are urban area, roads, forest, 
agricultural area, sea, lake and sand. Classification of the 28 
June 2007 is shown in figure 4 as the third image of the 
classification. This classification besides cloud classification