Temporal change detection by principal component transformation on satellite imagery

The principal components transformation is used to detect temporal changes in Istanbul using Landsat MSS, TM and Spot XS satellite imagery acquired respectively in 1975, 1984 and 1993. This investigation employs the feature orientated principal component selection property of the KL transform to emphasize the temporal changes of the last two decades. It is shown that application of principal components transformation on temporal data of the study area emphasizes temporal changes in the least significant information content components which are usually considered as noise and discarded. The study area is that part of Istanbul in Thrace consisting of a 20×20 km² area bound in the south by the Golden Horn and in the north by the Black Sea where the shore is degraded due to mining activities and there has been considerable temporal changes in the green forest areas and at a small dam. Temporal changes due to construction of Fatih Sultan Mehmet bridge in 1992 on the Bosphorus and a part of its beltway are included in the changes detected by principal components transform