Abstract
The unsupervised classification was used for the data (1989, 2009) of the American satellite Landsat. The result showed that we obtained six land cover types (dense forests, open forests, pastures, agricultural lands, soil and rocky lands). Were These all lands types identified and compared with the land control points of (65) samples and for both of the data. The accuracy of classification of the spatial data was calculated and evaluated depending on the scales used for this type of studies, which include: the error matrix of the percentage of each type and for the map as a whole. The accuracy of both of the data (1989, 2009) were (84.62%) and (83.08%) respectively and that indicates a good accuracy for both classifications. Also the statistical scale (Kappa) was used to calculate the accuracy which was (0.81 and 0.79) respectively for the two data classified for the year (1989 and 2009). The change in the covers in the period between the data collection showed that there were increases in the open forests, pastures and the agricultural lands with percentages of (2.924%, 3.78% and 1.225%) respectively. The high increase was in the open forests and pastures. But the increase was slight in the total area of the agricultural lands. From the other hand, we see a loss in the area of the dense forests, soil and the rocky lands. The highest loss was in the dense forests (2.242%) and rocky lands (1.918%), while the loss in the soil was slight (1.769%).
