Developing Decision Tree and Data Mining Based Conceptual Model for Detecting Land Cover Changes Using TM Images and Ancillary Data

Rapid urban growth and industrialization have caused many environmental problems in a number of cities around the world. Knowledge about land cover/land use changes in the long term is very important for urban managers and policy makers in order to evaluate and predict the resulting problems. Remote sensing is an effective tool for monitoring these changes in urban areas and its periphery. Over recent decades، Yasouj City has developed and affected its surrounding environment due to urban growth and immigration. The objective of this research is to develop a Decision Tree and data mining based conceptual model for land cover change detection using a Landsat Thematic Mapper (TM) and ancillary data in the central Section of Boyerahmad County from 1990 to 2009. Based on findings of the study، the overall six-class classification accuracies for 1990 and 2009 were، respectively، 93. 16% and 93. 37%. The overall accuracy of land cover change maps، generated from post-classification change detection methods and evaluated using two approaches، ranged from 85. 6% to 86. 98%. The maps also showed that between 1990 and 2009 the urban area increased by approximately 19. 28% while agriculture and forest decreased by 31. 76% and 7. 32% respectively.