MONITORING AND PREDICTION OF URBAN EXPANSION USING MULTILAYER PERCEPTRON NEURAL NETWORK BY REMOTE SENSING AND GIS TECHNOLOGIES: A CASE STUDY FROM ISTANBUL METROPOLITAN CITY

The most significant occurrence in this era is rapid urbanization. Especially in developing countries, the cities have an obvious expansion due to rapid population growth, economic development and infrastructure development initiatives. It is crucial to monitoring and to predict the urban expansion for sustainable cities in terms of settlement planning, transportation, landscaping, etc. At this point, remote sensing data (especially satellite data) and geographic information system (GIS) techniques allow us to get necessary data with high spatial, spectral and temporal resolution. This study aims to detect the Land Use and Land Cover (LULC) of the Istanbul Metropolitan City from 2003 to 2016 in purpose of urban expansion by using Landsat 5 and Landsat 8 satellite images in a GIS environment and to predict the status of the city in 2030. The study was carried out in two main steps. In the First step, the LULC changes are detected in Land Change Modeler (LCM) between initial and final dates. This step is very similar to a basic change detection process. The second step is to predict the city statues for a future date. In this step, the results from first step such as classification images are evaluated in a Multilayer Perceptron (MLP) approach and then the expansion of the city is predicted. In our case, the change rate of built-up area from 2003 to 2030 is predicted as (+) 48%. The MLP approach allows us to detect the transitions between LULC classes which are very important for city planning. According to results the main transition to built-up area will come from agricultural land. In the end, this study exhibits the advantages of remote sensing and GIS and the importance of urban expansion monitoring in terms of prediction.