Land-use Change Detection at the Rural-urban Fringe using Multi-sensor Data in Ankara, Turkey

要旨

Land-use changes at the urban fringe were detected using a combination of Landsat Thematic Mapper (TM) and IRS-1C PAN images of Batikent, Ankara, Turkey. Change detection was carried out in two stages. First, seven major land-use change classes were detected through a multi-date classification of Landsat TM images obtained in 1985 and 1995. To minimize the uncertainty, existing areas of urban land-use were masked out and eliminated from the change detection analyses. The overall classification accuracy was computed as 80.9%. The detailed urban land-use categories were then detected through visual analysis of the fused Landsat TM (1995) and IRS-1C PAN (1997) image. It is shown that a hybrid multi-date classification and visual analysis technique can be effective to detect rural-to-urban land-use change. Multi-date classification provided the general land-use change classes. The fused Landsat TM and IRS-1C PAN data was suitable for visual analysis of detailed urban land-use classes. The results demonstrate that the study area experienced a rapid rural-to-urban conversion in the 10-year period between 1985 and 1995.