Hyperspectral-based characterization of different vegetation species in Sundiken Mountain, Turkey

摘要

In this study, derivative spectroscopies of collected hyperspectral reflectance data of forest vegetation were used to obtain the wavelength peaks and troughs of different forest species of Sundiken, Eskisehir, Turkey. Dominant vegetation species at this location are Pinus brutia Ten, Pinus nigra Arnold, Pinus sylvestris and Quercus pubescens. Especially, the reflectance spectra of the different pine species have similar spectral features. So, it is very difficult to classify evergreen trees using the ordinary multispectral sensor. Hyperspectral reflectance of three type pine and one type oak species in the Sundiken mountain, Turkey, was measured using an ASD FieldSpec Pro FR spectrometer over the 350 to 2500 nm range and high-order derivative spectroscopy was applied to the hyperspectral reflectance data to discriminate four type plant species. Major peaks in the fourth derivatives indicate that wavelengths of around 670 nm, 690 nm and 730 nm are effective in discriminating of dominant forest species. In this investigation, high-order derivative analysis is shown applicable to hyperspectral reflectance spectra for separating out peaks of overlapping bands.