Effect of thermal bonding and supervised classification algorithms of satellite data in making land use maps (Case study: Kashan)

Today, remote sensing data can provide the latest information for the study of land cover and land uses These images are of utmost importance due to the providing updated information, variety of forms and the possibility of processing for making land use maps. Determining the location of each land use together with land cover helps managers in making decisions. Also, the current status of the land cover can be studied by using land use maps at different levels. In order to evaluate the ability of multitemporal satellite data of TM and ETM + in land use classification and the effect of the thermal band on increasing the accuracy of the land use maps, Landsat TM digital data relating to the September 27, 1990 and Landsat ETM + digital data of 10 July 2002 from Kashan were analyzed. Initial studies were performed on images in terms of the presence of geometric errors. In order to prepare the images for digital processing, improved operation and image enhancement were applied on the images. Ground truth map was prepared in the same stage. Then, the supervised classification of satellite images with algorithms and different approaches including a variety of bands were tested and the accuracy of each of the methods and approaches were investigated. The results showed that the highest Kappa accuracy in both Landsat TM and ETM + with all the bands were 86.34 and 83.21, respectively. While the elimination of the thermal band decreases the accuracy to 82.46 and 79.93 % The results of this study showed that using thermal bonding caused an increase of 4 percent in Kappa accuracy, and the highest accuracy was occurred in the classes of clay plains, flood plains, mountains and puffy salt lands. Therefore, despite the lower spectral resolution, the use of thermal bonding is recommended in such studies.