Hyperspectral Remote Sensing Estimation Model Of Heavy Metal Content In Wetland Plants Leaves

This article selects Shanyutan in Minjiang estuary wetland as the study area, gathering the blades of phragmites australis and Spatina alterniflora in this area as the experimental sample, collecting the spectroscopic data and measuring the assay of lead (Pb) and copper (Cu) in blades in laboratory. Using the correlation analysis method in extracting high spectral characteristic parameters, establish an estimation model in wetland vegetation blade heavy metal content of hyperspectral remote sensing by single variable regression and multiple stepwise regression method, the following main conclusions:(1) Found that Pb content and the correlation between variables are better than Cu content and the correlation between variables by the first derivative of spectral variables, variable continuum-removed spectrum, trilateral parameter, green peak parameters, Red Valley and paper build parameters and normalized difference vegetation index, homogenization greenness index, modified chlorophyll index, ratio vegetation index, the index structure effect, photochemical reflectance index.(2) Estimating the heavy metals content of phragmites australis, Pb content based on the first derivative variables dv822build a linear model fitting and forecast effect is relatively good, Cu content based on the original spectral variables R426build a linear model fitting and forecast effect is relatively good.(3)Estimating the heavy metals content of Spatina alterniflora, Pb content based on the first derivative variable dv945constructed the index of function of the model fitting and forecast effect is relatively good, Cu content based on the yellow area of the exponential function of Sy build model fitting and forecast effect is better.(4) Lead and Copper content are estimated in different kinds of wetland plants (i.e., phragmites australis and Spatina alterniflora mixed samples), which found that the quadratic function model based on variable PRI(Photochemical reflectance index) build for Pb and Cu content have good estimation results.(5) Getting the estimate and prediction of single plant by using the method and the estimation model of this paper is better than the two mixed vegetation, the reason may be due to the samples have some influence between each other.