Object- Based land use/land cover change detection using spatio -temporal images - A case study in metropolitan city of Istanbul, Turkey

Abstract

Due to the high industrialization and corresponding increase in population, a significant number of housing request was witnessed in the last several decades in metropolitan city of Istanbul, Turkey. As a result, a significant amount of land degradation and environmental change has occurred, particularly in the outskirts of the city. According to the Turkish Statistical Institute's (TUIK) report for 2018, Istanbul is the most populous city in Turkey, reaching to 15 million. For the rise of human needs, new transportation structures including roads, airports and bridges have been lately built in the city. Hence, the construction of built-up areas has led to an essential fluctuation in land use/land cover (LULC) in the region. The construction of the Istanbul Grand Airport (IGA) (2014-2018) and the Yavuz Sultan Selim Bridge (2012-2016) have affected the wetlands and forests directly affecting ecological habitat. In this study, two satellite images acquired in July 2013 and April 2018 from Landsat 8 OLI/TIR were employed to investigate the LULC changes in the city. In this context, six main LULC classes namely, water, road, bare soil, urban area, forest and agriculture were determined. The five-year period of change detection of city was performed by using object-based image analysis (OBIA) using multi-resolution segmentation algorithm. The LULC types from the created image segments were determined by using spectral indices, mean, minimum, maximum, standard deviation, brightness and maximum difference of spectral bands. In the classification phase, the nearest neighborhood classifier was applied for producing LULC thematic maps using the multi-temporal images. In parallel to the construction of the IGA (new Istanbul airport) and Yavuz Sultan Selim Bridge, a noteworthy increase in urbanization was observed. Particularly, the area of road class has vastly expanded on a spatial basis, almost doubled in size. Another important finding is that acreage of land water bodies decreased about 500 ha due to the construction of IGA on wetlands and marshlands. Results showed that the study site was subject to approximately 24% LULC change, and the highest change occurred between bare soil and forest classes, corresponding to 5% transition between these classes. © 2021 Elsevier B.V., All rights reserved.