Semeniv O. The identification methods and algorithms in monitoring and predicting vegetation state

The thesis deals with the problem of new approaches development for monitoring and prediction of vegetation state using methods of identification and optimization. The method of nonlinear discrete models identification for yield prediction is presented. The problem of dynamical systems reconstruction is described and it solution by optimization methods is represented. The new optimization model of risk minimization in prediction of vegetation production is performed. The implementation of the developed identification algorithm for geomagnetic Dst-index prediction is described. The problem of regression models reconstruction for vegetation biochemical content estimation using information characteristics of spectral data is described. The system analysis is used for the selection of the information characteristics of the reflectance spectra. The new linear and nonlinear regression models for chlorophyll concentration estimation in leaves are developed. New methods of vegetation state express-analysis are described. The information technology for the adaptive spectral data processing is justified on the example of the field spectrometer data processing. The unique software for the first Ukrainian field spectrometer is represented. The new model of device technical characteristics optimization is represented.