Biochemical Component Estimation Model Explore Based On The Spectral Characteristics Of The Forest Leaves

With the rapid development and constantly improve of hyper-spectral remote sensing technology, the early days of the birth of hyper-spectral remote sensing technology, which mainly be used in the field of geological survey. After1988, hyper-spectral remote sensing technology has been successfully used in atmospheric science、ecology、geology、hydrology and marine disciplines. Due to the high spectral is famous for extremely high resolution is extremely high and strong continuous band, it can obtain a fine spectral information of target, it also identify target directly and analyze differences of weak spectral.This article is based on theories, which are extracted from hyper-spectral remote sensing of vegetation physicochemical characteristics and agronomic parameters. Linear and non-linear regression analysis and stepwise multiple regression analysis method predicted chlorophyll、Carotenoids、carbon and nitrogen conten of Taiwan acacia, eucalyptus, and the Schima.In order to find some suitable wavebands and characteristic parameters to estimate biochemical component content of each tree leaves, and establish mathematical models between the relationship of chlorophyll, Carotenoids, carbon content, nitrogen content and parameters of spectral characteristic.This article mainly used three methods:Firstly, using different mutative forms of leaf spectrum and different biochemical parameters variable to build regression relationship, different mutative forms of leaf spectrum included the first derivative and the original spectrum; secondly, using the Trilateral position,width,wave deep the Trilateral area and leaves biochemical component parameters to build regression relationship; thirdly, inversion spectral curve through geometrical optics model, building relation on the basis of parameters extracted from sensitive bands of curve and biochemical components.The conclusions of this article include:(1) The process spectral data is much better than the original spectral data, processes include the first derivative, continuum removed and envelope removal. It is great potential to use "trilateral" parameter, band depth, wave deep center normalized, band area normalization variables to estimate biochemical component content(2) In the construction of estimating biochemical component content model, part middle infrared band except outside the range of visible and near-infrared region, there are a few bands play a certain synergistic effect in the estimation of chlorophyll and carotenoid content.In the estimation model of biochemical component of three species, estimation model of chlorophyll content and carotenoid content is much better than estimation model of carbon content and nitrogen content. This conclusion confirmed:it is necessary to combine spectral vegetation indexes and other physical parameters to estimate carbon content and nitrogen content.(3) based on the LIBERTY radiative transfer model and modulation and change of model parameters,the model can simulate leaf spectral reflectance curves under different chlorophyll content,different moisture content and mesophyll structure parameter.We fit average leaf spectral reflectance from visible to near-infrared bands estimate chlorophyll content by model.We also compared the model fitting chlorophyll with laboratory measured chlorophyll content.If accuracy of the model is higer, the same precision data with laboratory chemical methods can be obtained through estimated model.(4) Through analyzing results simulated by PROSPECT model,we found that changes of chlorophyll content chief influence the visible and shortwave infrared spectrum curves between400-900nm. Making use of this model can better simulate leaf spectral reflectance curve, the results of the sensitivity analysis of the different bands shows it is feasible to inversion leaf chlorophyll content by changing the variable input parameters and fixed parameters and a lookup table built by PROSPECT model, results and accuracy of inversion are higher, which can achieve the same accuracy with the laboratory measured data.(5) The accuracy of the model is higher, by the spectral curve under blade level, which estimates leaf biochemical component content by the spectral curve under blade level. It can meet the estimates of forest biochemical content, and then further provides research basis for retrieval tree canopy biomass and LAI by Hyperion image.