Investigation of Quickbird Satellite Image Capability in the Separation of the Canopy of Zagros Forest Trees

Aims 2005 DashteBarm forests of Fars province image is used to investigate and evaluate the capability of Quickbird satellite imagery to differentiate tree canopies regions from no-tree areas. Materials and Methods First, the validity geometric correction of satellite image was assured. By systematic random sampling, 79 square footages of (20*20) in ARCGIS 9.3 software was designed and on the footages’ places of the combined image from Quickbird panchromatic band and multispectral band, the samples of no tree canopies and tree canopies areas was obtained. Then, 20% of the footages were considered as test samples and the rest was studied as training samples. In the next step, processes on a multivariate image were performed by ENVI 4.3 software and some indexes such as NDVI, GNDVI, RVI Partial Component Analysis (PCA) were created and integrated and were combined. Then, two classifications on the original image and processed bands with two methods of maximum likelihood and Support Vector Machine (SVM) were categorized, in which the images were classified into two classes of trees and non-trees. Findings Evaluation of the classified images using the test samples showed that the accuracy and Kappa coefficient in the classified images of the original bands were 94.478% and 0.789 for the maximum likelihood method and 94.848% and 0.877 for the support vector machine, respectively. Also, the results of the processed band’s classifications by maximum likelihood and support vector machine methods showed that these images have 94.274 and 94.683% accuracy and Kappa coefficient of 0.875 and 0.882, respectively. Conclusion: The results of this study show that the Quickbird satellite image is suitable for separating tree canopies and no tree canopies areas in Zagros forests and similar areas.