Ukrainian Journal of Forest and Wood Science

Application of seasonal series of Landsat satellite images facilitates development of methods used for investigation of current state and dynamics of forests. An information on seasonal variability of spectral reflectance of different land cover types increases an accuracy of forest mapping. The article summarizes the analysis of temporal dynamics of spectral features collected for different landscapes of flat part of Ukraine. We used 621 scenes of Landsat 8 OLI acquired during 2014-2016. The spectral data were analyzed after radiometric correction of satellite images to TOA reflectance, cloud filtering and composing into 12 monthly mosaics. For the analysis we selected 6 bands of visible and infrared spectra as well as different combinations of these bands. Using stratified random sampling we extracted median values of reflectance for each land cover type and month. High resolution satellite images from Google Earth and Bings Maps were used for visual interpretation of sampling points. As a result, we concluded the good separability of land covers in shortwave-infrared range (Band 6 and Band 7) and following band combinations: Band 6 / Band 7 and Band 5 / Band 7. The difference in tree species composition of forest stands is the biggest for red and near-infrared spectra (Band 4 and Band 5), and in shortwave-infrared bands (Band 6 and Band 7). The reparability between tree species proups is negligible for both thermal bands of Landsat 8 OLI images. The conducted research forms the basis for selecting optimal combination of predictor variables for classification of seasonal composited mosaics using Landsat time series.

Landsat 8 OLI, seasonal composited mosaics, satellite images, spectral reflectance, spectral band