Comparison of three classification methods in vegetation mapping using remote sensing (Case study: shalmanrud Basin)

Map information in research and studies is of great importance. The satellites are very suitable for the assessment of natural disasters because they have an extensive and integrated view, with a large part of the electromagnetic spectrum and up-to-date. The present study evaluates three classification (unsupervised, supervised and hybrid) methods for mapping vegetation in the Shalmanrud region. Using the ETM2002 satellite imagery, the 166-channel Land-Sat Satellite 34 in the ILWIS software version 3.1 was compiled,After performing the necessary corrections and initial processing, data classify was performed. Finally evaluated, efficiency of classification methods using index User accuracy overall accuracy and kappa coefficient. After checking the numbers obtained the index Comparison of three classification methods with the ground truth map showed that the supervised classification method using the Maximum likelihood method with a total accuracy of 67.84% and kappa coefficient of 0.6752 had better results than the other two methods. Also, analysis of the accuracy of each classification and comparison with ground truth map showed that the supervised classification method with a precision of 75.14% has the best result compared to the other two methods for the studied area.