Inventory of Single Oak Trees Using Object- Based Method on WorldView-2 Satellite Images and on Earth


1 PhD student of forestry in Ilam university

2 Associate Professor and Faculty Member of Forest Sciences Department in University of Ilam

3 Associate Professor, Department of Natural Resources and Environment, College of Agriculture, Shiraz University, Shiraz, Iran


Remote sensing provides data types and useful resources for forest mapping. Today,
one of the most commonly used application in forestry is the identification of single
tree and tree species compassion using object-based analysis and classification of
satellite or aerial images. Forest data, which is derived from remote sensing methods,
mainly focuses on the mass i.e. parts of the forest that are largely homogeneous, in
particular, interconnected) and plot-level data. Haft-Barm Lake is the case study which
is located in Fars province, representing closed forest in which oak is the valuable
species. High Resolution Satellite Imagery of WV-2 has been used in this study. In
this study, A UAV equipped with a compact digital camera has been used calibrated
and modified to record not only the visual but also the near infrared reflection (NIR) of
possibly infested oaks. The present study evaluated the estimation of forest parameters
by focusing on single tree extraction using Object-Based method of classification with
a complex matrix evaluation and AUC method with the help of the 4th UAV phantom
bird image in two distinct regions. The object-based classification has the highest and
best accuracy in estimating single-tree parameters. Object-Based classification method
is a useful method to identify Oak tree Zagros Mountains forest. This study confirms
that using WV-2 data one can extract the parameters of single trees in the forest. An overall Kappa Index of Agreement (KIA) of 0.97 and 0.96 for each study site has been achieved. It is also concluded that while UAV has the potential to provide flexible and feasible solutions for forest mapping, some issues related to image quality still need to be addressed in order to improve the classification performance.