In recent years, remote sensing data and geographic information system (GIS) have been widely applied in identifying and analyzing land use and land cover change. Accurate change detection of earth's surfaces provides better understanding of relationships and interactions between human and natural phenomena for prefect resource management. There are a lot of methods for land cover change detection such as image rationing, image difference, change vector analysis, image regression, composite analysis and post- Isfahan land cover from 1951 to 2006. Aerial photos at a scale of 1:50000 in 1955 and MSS, TM and ETM + images of LandSat satellite in 1972, 1990 and 2001 respectively were used in the current study to produce land cover maps and study trend of changes in various years. The predictive land cover map by CA-Markov for 2006 was used for completing the study period. For studying this research all of the aerial photos and satellite images were rectified by using first-degree polynomial equation and nearest neighbor sampling. Aerial photos were interpreted visually by using color, tone, pattern, shape, location and various phenomena of area. Each Landsat image was enhanced using linear contrast stretching and histogram equalization to improve the image. To make the best false color composite, we used Optimal Index Factor (OIF) to identify best bands with minimum correlation and maximum variance. RMSE of aerial photos was in a range of 0.1 to 0.45 pixels. The resultant root mean squared error of MSS, TM and ETM + images were found 0.73, 0.68 and 0.6 pixel, respectively. To 0.05. By performing goodness-fit test and calculating ? 2 and comparing it with ? 2 in the table at ?= 0.05 with freedom degree of 4, null hypothesis was rejected and One hypothesis (mean difference in the results from the two mentioned methods) was confirmed. Calculated ? 2 from predicted map of years 1990 and 2001 are 25642 and 36915, respectively that at ?=0.05 and freedom degree of 4, ? 2 is 0.711 which rejects null hypothesis. To calculate kappa coefficient and overall accuracy of all land cover maps, ground control points were used; then error matrix was calculated for producing land cover maps by using Hybrid method and CA-Markov model.
