SINOPSIS

Consideration is given to the problem of natural ecosystems state estimation using satellite observations. Original procedure of remote sensing data processing and natural ecosystem complexes rapid analysis are presented. Its main component is the classification model received with the use of the statistical learning methods. As an example the results of processing and classification of Svityaz lake satellite images between 1988 and 2009 are presented. The results of the model testing are presented, the problem of finding the optimal model with high generalized capability and classification accuracy is studied. The numerical algorithm is developed for rapid analysis of natural water ecosystems space images which allows one to obtain the classification accuracy of more than 95%.