M.Farahnak Ghazani, M. Najibzadeh
Rangeland Research Division, Research Institute of Forests and Rangelands, Tehran, Iran
Research Center for Agriculture and Natural Resources, Tabriz, Iran
Key words: Forest, ETM+, classification, accuracy, Arasbaran.
Current research was carried out in Arasbaran protected area in order to investigate the capability of landsat ETM+ satellite images in forest cover mapping. For this perpose the ETM+ image of the study area acquired on june 16th, 2001 was used. Radiometric and geometric correction were done for the images. The best spectral bands were selected using Optimum Index Factor (OIF) method. False color composite map was created using bands 4, 3, 2 as the best combination of spectral bands. Sufficient number of training samples of each land cover class were collected for image classification. Supervised classification was accomplished using maximum likelihood classifier and forest cover map of the study area was created. According to the results the area of forest and non-forest classes are 20777.7 ha and 59813.9 ha respectively. Ground truth data was collected using Systematic-cluster sampling method In order to assess the accuracy of classification. The results showed that overal accuracy and kappa statistic are %95 and %88 respectively.
Get the original articles in Source: Volume 4, Number 2, February 2014 – JBES
Published By: International Network for Natural Sciences