Investigation of the Expansion Trend and Related Effective Parameters Using Artificial Neural Network (Case Study: Tabriz)

Abstract

Prediction of urban growth and the encroachment of urban construction are very important. The planning of city managers for future management of the city and even planning to invest citizens are such cases that can be referred to. In this research, the trend of changes in urban development has been done in Tabriz, using artificial neural networks (ANN). To implement this, multi-temporal Landsat TM images for years 1990, 2000 and 2010 were used. To classify the mentioned image, MLC algorithm was used and post-classification comparison method was implemented for the change detection process. Feed-forward network architecture along with Back propagation algorithm has been used to correct the network weight during the process. Input variables include distance from the main roads, established areas, slope, service areas and aspect layers. The role of geological and tectonically parameters was also considered. Accuracy assessment of the simulation was performed for year 2010. In this process, the output of the neural network as forecasted image was compared to classified image of the same year. It shows the accuracy of 96%. The results show an increasing urban area growth via destroying vegetation as well as agricultural areas, in such rate that, it is forecasted more than 16000 hectares of nonurban areas will be transformed to urban areas during years 2010-2030. Also the results of this research indicate that the expansion direction of the city for the mentioned period will be toward the southeast.

Keywords

• artificial neural network
• simulation
• urban growth
• gis
• tabriz