عنوان مقاله [English]
The present article proposes land transformation model which consists of Geographic Information System (GIS) and Artificial Neural Networks (ANNs). This model applies varied political, social and environmental models as predictive variables. The study introduces a version of LTM model for Grand Traverse basin in Michigan gulf and shows how factors like roads, highways, and local streets, and rivers, coastlines in large lakes, entertainment facilities, inland lakes, agriculture density and landscape quality can affect urbanization pattern in coastal basin. GIS is used for understanding local patterns of development, estimating predicting capacity of the model from artificial neural network, spatial expansion of predicting stimulators, and spatial analysis. Finally, the contribution of each predicting variable is estimated and presented on a spatial scale. Landscape quality was the strongest predicting variable on the smallest scale. Multi-scale impacts of land use changes are analyzed using the relational impacts of the site (like landscape quality, local streets) and position (like highways and roads between different regions) on different scales.