Investigation Of The Biochemical Components Content In Trees Leaves Using A Hyperspectral Remote Sensing Estimation Model

With the rapid development and continuous improvement of hyperspectral remote sensing technology, it has been widely used in the geological, environment and vegetation remote sensing monitoring. Due to the high spectral resolution and strong waveband continuity, it can obtain the refined target spectral information, laid a foundation for direct target recognition and faint difference analysis of spectrum.This study is based on the theoretical guidance of hyperspectral remote sensing technology to the vegetation physicochemical properties and the agronomy parameters extraction research. Adopted the correlation analysis, linear and nonlinear regression and stepwise multiple regression method to estimate the leaves chlorophyll, carotenoids, C and N content of Taiwan acacia, eucalyptus and schima superba. The purpose of this study was to find some sensitive wavebands and spectra characteristics parameters which were suitable for estimating the leaves biochemical component of chlorophyll, carotenoids, C and N content, then establish the relation model between the biochemical components content and spectra characteristics parameters.The main contents of the study as following:Collecting the leaves spectrum of Taiwan acacia, eucalyptus and schima superba, then determination the leaves chlorophyll, carotenoids, C and N content respectively, and then,(1) Using the spectrum analysis technique and statistical method to analysis the spectral data, extraction the sensitive wavebands and spectra characteristics parameters with the leaves biochemical components content;(2) Adopt the single variable regression and stepwise multiple regression to establish the leaves biochemical components content estimation model with the spectra characteristics varible;(3) Evaluate the accuracy of each estimation model and the potential of each spectra characteristics varible to estimate the leaves biochemical components content.