LAND USE MAPPING USING VISUAL AND DIGITAL INTERPRETATION OF TM AND GOOGLE EARTH IMAGES IN SHIRVANDARASI WATERSHED (NORTH-WEST OF IRAN)

摘要

This study was conducted to use Landsat and Google earth (digital globe) imagery for land use mapping in ShirvanDarasi watershed in north of Ardabil province. A TM image by considering seasonality and phenological pattern was selected. Pre image processing stages such as atmospheric and geometric correction, and topographic normalization were conducted before image utilization. Moreover, image of the study area extracted from Google earth and imported to ArcGIS environment. Ancillary data such as DEM and slope were derived and added to the datasets of this study for controlling different land uses. Field visit and appropriate ground control points were collected for visual and training area selection, and finally land uses such as rangeland, orchards, irrigated and dry farming, residential and industrial areas, roads and out crops were considered and land use of the selected images were derived. Finally accuracy of the produced maps were computed and compared. Results show that, the produced map of the image of Google earth using visual interpretation showed high overall accuracy (90%) and Kappa (0.94). On the other hand, results of the digital interpretation of TM image (unsupervised) showed very low overall accuracy (24%) and Kappa (0.24) statistics.