EN
Evaluation of the forest landscape diversity was investigated based on the multispectral aerial images using iterative Principal Component Analysis (PCA) methodology. In 2014, we carried out several photogrammetric flights over the experimental plots establish in the Krotoszyn Plateau (central Poland) documenting the vegetation cycle of forest stands dominated by oaks. Aerial photos of the spatial resolution about 25 cm of forest area in Karczma Borowa Forest District in the range of visible light (460−650 nm) and near infrared (700−930 nm) were collected by multispectral Quercus 6 platform placed on the aircraft. The aim of the study was to evaluate the diversity of forest vegetation cover using remote sensing data based on spectral signatures of plants without complete classification of fractional vegetation cover and species identification in the field. Recursive PCA on data collection from the multispectral images helped to determine with the semi−automatic mode the number of land cover classes, including the classes of vegetation. Based on the radiometric data, the separation of inorganic matter from vegetation and diversity indicators of forest stands on the image area were evaluated. With the PCA method, along the most volatile vectors, the first division into land cover classes of vegetation was conducted. As a result of the first iteration of PCA, three classes of vegetation: deciduous trees, conifers and forest undergrowth was determined. In the second iteration, classes of forest vegetation were separated and interpreted as the area dominated by a single species of tree or shrub. The second iteration divided the deciduous plant image area in plots dominated by English oak stands with an admixture of birch and red oak. Based on the number of pixels in classes representing individual plant species, Shannon−Wiener (H) and Simpson (D) diversity indices were determined. By described methodology, it was found that the differences between the H and D indices for the imagery area after the first and second PCA iteration were small. The relevance of performing successive iterations of PCA analysis, and thus the full identification of species, in the context of diversity calculation should be the subject of further study.