Evaluation of Landsat-ETM+ and IRS-LISS III satellite data for forest type mapping in Zagros forests (Case study: Ghalajeh forest, Kermanshah province)

• Main Authors: Rohollah Porma, Sha'ban Shataee Joybari, Yahya Khodakarami, Hashem Habashi
• Format: Article
• Language: fas
• Published: Research Institute of Forests and Rangelands of Iran 2009-12-01
• Series: تحقیقات جنگل و صنوبر ایران
• Subjects: ETM+; LISS III; maximum likelihood; forest type mapping; sampling ground truth; Zagros
• Online Access: http://ijfpr.areeo.ac.ir/article_107760_ce46c6ae4ff68e84977d4e72ffc3940e.pdf

In order to evaluate and compare the capability of ETM+ and LISS III data for forest type mapping in the Zagros forests, a small window of panchromatic and multispectral images of Landsat-ETM+ and IRS-P6-LISS III satellite data were selected from Ghalajeh forests in the Kermanshah province. No radiometric error was found using the quality investigations. Orthorectification of ETM+ was done using 55 ground control points with RMS error of 0.39 for X axis and 0.46 for Y axis and for LISS-III imagery with 34 ground control points with RMS error of 0.67 for X axis and 0.58 for Y axis. Some suitable image processing functions such as principal component analysis, tasseled cap transformation and appropriate vegetation indexes were applied for classification processes. In order to assess the classification results, a sample ground truth was generated using a systematic network with 60m×60m sample area. By computing the canopy cover percent of species, four forest types were determined in the study area. By selecting 25% of samples for each class as training samples, the best band sets were selected using transformed divergence separability index. Classification was performed by supervised method using minimum distance (MD), maximum likelihood and parallel epiped (PPD) classifiers. Results of classification showed that overall accuracy and kappa coefficient for 5 classes for ETM+ images were obtained %44.57 and 0.18 and for LISS III Images %50.6 and 0.32, respectively. After merging the classes of 1 and 2 due to spectral overlapping, the overall accuracy and kappa coefficient for 4 classes using ETM+ images were obtained %61.08 and 0.21 and for LISS III Images, %71.44 and 0.33, respectively. Finally, by merging the classes of 3, 4 and 5, classification was done with two types and the overall accuracy and kappa coefficient obtained %74.1 and 0.37 for ETM+ and %77.7 and 0.41 for LISS III, respectively. Being open canopy cover as well as conflicts between soil and vegetation reflectance caused preventive of obtaining the more favorite results. Result showed fairly more capability of LISS III data in compare to ETM+. Similar research in other regions and using of higher multispectral resolution data is suggested