Musaoglu, Nebiye
national forest inventory and monitoring of the periodical changes in forest areas, are combined with aerial photographs
to be used in preparing the plans of forest management ( Hame and Rauste, 1995).
By applying classification algorithms to satellite images of different dates belonging to Sariyer township, temporal
changes of the area were analyzed in this study.
2 STUDY AREA AND TEST SITE
In order to evaluate multitemporal data sets, Sariyer region located on the western part of Istanbul has been chosen
(Fig. 1). Sariyer township lies between the co-ordinates of 28° 61' 25" - 28° 61' 25" east, 41? 16' 37"- 41° 04' 55" north
and covers an area of approximately 151 square kilometres.
While the total area of Sariyer was covered with forests in ancient times today only the northwest side has remained as a
forest. In the forest areas, the vegetation is formed by natural species of Fagus oriantalis, Castania sativa, Quercus
spp., Carpinus betulus, Tilia argantena as well as Pinus nigra, Pinus maritima, Pinus pinea and other coniferous
species that have been planted, which exist pure or mixed in the forest. Apart from this, shrubbery also flourished in the
forest lands. Coniferous species grown by planting generally exist in the north east of the area
(Musaoglu and Ormeci, 1999).
Study Area
Test Area
Figure 1. Study Area
2.1 Test Site
In classification, Education and Research Forest of Istanbul University was chosen as the test site. This test site which
covers and area of 739.13 hectars, consists of pure and mixed tree species of all ages , especially mixed tree stands such
as mainly Castanea sativa(Cs), Pinus nigra(Pn),Carpinus betulus(Cb), Qercus spp(Qs), Tilia argentena(ta) as well as
Sorbus domestica(Sd), Fraxinus(F), Arbutus unedo(Au),Cistus spp(Cs),Sorbus torminalis(St),Spartium junceum(Sj)
and Laurus nobilis.(Ln) (Musaoólu, 1999).
In choosing the test site for classification, tree stand map at 1/5000 scale was taken as the basis. In preparing this map,
differentiation of the tree stand species was determined by considering the general characteristics of inaccessible areas
due to the thick forest lands and the topography of the area, and it was obtained by combining the species according to
density of the tree species. Tree stands map of the test site was digitized and transferred to computer media. Each tree
stand was taken as one layer in digitizing (Fig.2).
940 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.
WI