Abstract :
In order to assess and compare the capability of ETM+ and LISS-III data in the mapping of riparian forests of the Maroon River in Behbahan, Iran, a small window of panchromatic and multispectral images of Landsat-ETM+ and IRS-1D-LISS-III satellite data was selected. The quality of data and radiometric error was checked and geometric correction was implemented using 25 ground control points. Principal component analysis (PCA), tasseled cap transformation and appropriate vegetation indexes (NDVI) were applied to provide the main bands for incorporation in the classification processes. The study also used image fusion by Pansharp (the Gram-Schmidt Spectral Sharpening method) and the HSV method for high spectral and spatial resolution. After selecting the training data, the classification was carried out by choosing seven-class and three-class land-use/cover using a maximum likelihood algorithm. Transformed divergence and Jeffreys–Matusita were used to test the separability of the classes. Considering the results, it can be concluded that IRS1D-LISS-III and landsat 7-ETM+ data have a suitable ability for mapping the Maroon riparian forest as well as the classification of forest to separate landuse. The classification using original bands of ETM+ is better than LISS-III data due to higher spectral resolution. The opposite of these results is only shown in the classification with three user classes. The overall accuracy and kappa coefficient in this method were 98.80 and 0.84, respectively.
Keyword :
Behbahan, Riparian forest mapping, LISS-III, ETM+, maximum likelihood.