Hyperspectral Estimation Model Of Pigment Content In Al.pine Grassland

With the increase of the information data,range of band and training samples,traditional remote sensing technology can not realize the detection of more information of ground objects,hyperspectral remote sensing has emerged as the times require.Hyperspectral remote sensing is an important technology to obtain information of ground objects.It can provide more details of ground spectral details,so that hyperspectral data can be used to refine the classification and recognition of ground objects.Hyperspectral remote sensing has been applied to vegetation classification and identification,and plant biochemical parameters estimation.At present,the studies are mostly applied to agriculture and forestry.The research is relatively mature based on the analysis of spectral characteristics and biochemical components of crops and tree species.For the meadow,the correlation analysis of hyperspectral features is relatively few,and is at the initial stage.In recent years,the grassland in the plateau has been degraded and even desertification in the world.As the main carbon source system of the earth,the grassland ecosystem is closely related to the climate change of the earth.Therefore,it is necessary to carry out the research and analysis based on hyperspectral technology for the alpine meadow.This study is based on the meadow dominant species of Stipa purpurea in Shenzha County,Tibet.The spectral reflectance of Stipa purpurea was obtained by using portable spectral radiometer SVC HR1024(spectral range 350~2500nm and channel number 1024).The content of kinds of photosynthetic pigment in Stipa purpurea was determined.Based on the measured vegetation hyperspectral data and photosynthetic pigments content,we analyzed the reflectance spectrum characteristics of Stipa purpurea.By using the EXCEL data anal.ysis tool and the SPSS22.0 statistical software,the correlation anal.ysis of the pigment content of Stipa purpurea with hyperspectral.reflectance and its transformation parameters were carried out by multiple linear stepwise regression analysis and curve fitting.The characteristic parameters of photosynthetic pigments were selected,and the estimation model was established.Then three kinds of precision test indexes were used to evaluate the prediction accuracy of multiple estimation models of photosynthetic pigments.Finally,the optimal hyperspectral estimation model was selected for the estimation of photo photosynthetic pigments of the Stipa purpurea.This study provides the basic data for the study of the degradation mechanism of the al.pine grassland in the northern Tibet,and provides a theoretical basis for the maintenance of the healthy and sustainable development of the alpine meadow.The results of this study are listed as follows.(1)Through the analysis of the existence of Stipa purpurea,the altitude and soil moisture in the alpine meadow affect the breeding status of Stipa.With the elevation of al.titude,the soil humidity is low and the breeding efficiency is lower,but the growth of grass plants needs enough nutrition to make the growth of the plant better.(2)Based on the indoor measurement of photosynthetic pigment content in the laboratory,the content of chlorophyll a of Stipa purpurea was about 4 times the content of chlorophyll b,while the carotenoid content was generally a little more than chlorophyll b,so the chlorophyll a was dominant in the absorption.The correlation between chlorophyll a,chlorophyll b and total chlorophyll is better,and the correlation between carotenoids and chlorophyll is poor based on the correlation analysis of all kinds of photosynthetic pigments.(3)Based on the primary hyperspectral reflectance and its various data forms(spectral location,spectral area,spectral area ratio)and vegetation index,the correlation between the photosynthetic pigment content and the correlation of photosynthetic pigments is Chlb>Chla>Chls> Cars,and the correlation between the carotenoid and the carotenoid is weak.(4)The spectral characteristic parameters of the pigments related to the Stipa purpurea were selected as the independent variable,the pigment content of the Stipa purpurea was the dependent variable,and ten functional models were selected to establish the hyperspectral remote sensing estimation model for the photosynthetic pigments.The model was tested and analyzed with the mean square root error(RMSE)and relative error(RE).Finally,the optimal hyperspectral remote sensing model for chlorophyll a,chlorophyll b and total chlorophyll content of the Stipa purpurea was screened.