Urban expansion modeling using medium resolution satellite imagery based on cellular automata (Case study: Tehran city)

Modeling of urban growth and changes in land use are an integral part of planning for sustainable development. The present research intends to model growth of the Tehran metropolis city in temporal and spatial distribution. Therefore, using temporal Landsat satellite images, land use maps for the years 1988, 2002 and 201 object-based approach have been classified. The next step for this two periods, 1988-2002 and 2002-2013 land cover changes, In terms of quantity, trends and changes in spatial distribution was analyzed. For this purpose, first, using logistic regression model the effective variables in relation to urban growth, included fourteen variables in both the natural - environmental and economic - social variables in the period 1988 to 2002 studied and quantitative effects of those was calculated and the map of urban growth potential was produced. Evaluation of logistic regression Performance using two indexes Pseudo R2 and ROC with respectively values 0.32 and 0.89, indicating a good fit of the regression and the appropriate descriptive capabilities. In the following, area of changes for the expected year using Markov chain analysis was quantitatively calculated. Finally, the output of the logistic regression model and Markov chain analysis, using the cellular Automata model, urban growth for 2013 was also modeled. Comparison 2013’s simulated image with 2013’s classified image shows used model with 93 percent accuracy for area and Kappa coefficient 0.87 has been a successful model. Accordingly the same model was used for estimating urban growth in 2025 by using classified images of 2002 and 2013.