Using the wavelet method for a multispectral remote sensing study of plant community spatial processes in broad range of spatial scales

For the analysis of landscapes, the method of multispectral image (MSI) representation as a uniform space-morphological hierarchy of the segmented images is proposed, convenient for their direct interpretation by a person and for the classification on the basis of expert knowledge. The basis of the proposed segmentation method is the hierarchical clustering (HC) approach in a combination with the wavelet transform. In the proposed method the wavelet transform is used for the solving of two problems: Preparation of data for convenient HC and enabling the execution of MSI segmentation. To present MSI in a form, convenient for recognition by generalization of a spatial and morphological structure in a uniform hierarchy, it is shown by examples, that the proposed method gives the expert an opportunity to successfully interpret MSI on the basis of brightness polyfeatures, using expert knowledge. This result may be useful for different geobotanical, syntaxonomical and ecological interpretations. Analysis of this type of representation has a large potential for yielding information about the influence of different factors of the environment