Assessment of Land Cover Changes in Arid Lands using ASTER and ETM+ Sensors (Case Study: Marvast Plain, Yazd Province)

Land cover is defined as a physical and biological cover of the earth’s surface that contains water, vegetation and bare soil. Limiting factors in arid areas such as water shortage and soil salinity are important factors to study the occurred changes in land cover. Due to the large expanse of land, change detection by the traditional methods is not sufficient and efficient; therefore, using of new methods such as remote sensing technology is necessary and vital. The overall aim of this study was to assess changes in land cover in arid lands of Marvast plain, located in south of Yazd. In this study, ASTER (Terra) data for summer month of July for years of 2003 and 2010 as well as ETM+ data (Landsat) for July 2002 and 2010 was used. In the first step, after pre-processing operations such as geometric correction, land cover maps of the study area were classified into 5 classes (agricultural land, rangeland, bare land, desert and wetland) using Maximum Likelihood method in both satellite types. Then, taking ground truth data, the classified maps accuracy were assessed by calculating the Kappa coefficient and overall accuracy. The results of ETM+ classification show that the areas of agricultural land and wetland were increased and decreased, respectively in time span of 2002-2010. The same changes were also observed in change classification maps of ASTER for the period of 2003-2010. In this time period, the area of wetland is decreased and changed into desert area. In addition, rangeland area has reduced and changed into agricultural land. The results show that produced maps of both sensors have accuracy of above 85% which reflects high performance of both images in land cover change analysis. In spite of the fact that ASTER data is a little more accurate, results of using Landsat data which has more accessibility and lower cost, for change detection analysis is also acceptable and valuable.