عنوان مقاله [English]
The present study is focused on the evaluation of FCD (Forest Canopy Density model) for the estimation of forest density in northern forests of Iran using Landsat 7 satellite data. The model was developed as a semi-expert system in Asia-Pacific region that estimates forest canopy density without any training areas. In this study, an ETM+ image dated July 18th, 2000 was analyzed. After preprocessing the satellite image, four basic indices of FCD model (Vegetation Index, Bare Soil Index, Shadow Index and Thermal Index), were calculated. Vegetation Density Index and Advanced Shadow Index were then calculated and forest density map (derived from FCD model) was finally extracted. The forest density map was classified according to the form presented by Forest, Range & Watershed Management Organization of Iran (7 classes), and another form (5 classes). In order to assess the accuracy of classified forest density map, a ground truth map of the entire study area was generated using aerial photos - at the scale of 1:10000 dated August, 1999. In this way, at first, the geometric correction of digital photos was implemented and the mosaic of photos was generated. Then, the Ground truth map was produced using on-screen digitizing method based on visual interpretation and applying stereoscope and printed photos. In this study, the highest overall accuracy and kappa coefficient - were obtained in classification in five classes - equal to 61% and 0.38, respectively. The overall accuracy and kappa coefficient in case of classification in 7 classes were less than those calculated in case of classification in 5 classes. This is because of the spectral similarity among the lower density classes. Hence, it could be concluded that in such forests, the potential of the model in separating high density forests, was relatively acceptable whereas the model could not correctly separate the lower density classes.