EN
The mainstream of remotely sensed methodology for identifying the tree stand condition is based on spectral responses registered by a multispectral sensor as a digital image. The changes in spectral properties are caused by dying leaves, needles or whole trees. In further steps, the relationship between the spectral values (radiometry) registered in a multispectral satellite image and the health condition of trees should be determined. The most frequent situation includes the one whem dying stand (sensu single tree) occupies the area of <5 m². Therefore the remotely sensed data for determining sanitary conditions of trees must be of a very high spatial resolution (e.g. WorldView2 or 3, GeoEye−1, Pleiades) on one hand and at the same time favourable for the vegetation studies, i.e. utilizing suitable spectral bands and be of low acquisition cost (e.g. RapidEye, LANDSAT−7, ETM +, LANDSAT−8 OLI). Thus a compromise between spatial and spectral resolution should be found to answer the question at what resolution it is possible to clearly separate the damaged tree. The scope of the research included testing of selected methods of satellite image processing and analysis in terms of defining the optimal spatial resolution, which was performed on simulated images obtained for the area of the Beskidy Mountains (S Poland). Pixel size on simulated images was downgraded to the size corresponding to the currently functioning satellite systems. Consequently the obtained material for comparison was free from influence of external factors such as the differences in: time and weather conditions, the geometry of satellite image acquisition, light at the surface of the treetops and phenological vegetation. For each image we used vegetation indices (NDVI and GDVI) and supervised classification. These tests and the obtained results allowed to draw conclusions about the optimal satellite image resolution that can be used to detect damaged or dead stands.