Research Of Crop Canopy Characteristic Information By Using Hyperspectral Red Edge Parameters Of Cotton

In this study, we select cotton as our interesting object, and use ASD FieldSpec Pro FR? to obtain the spectral reflectance of cotton .We carried field-spectral observations at ccotton different growth period and obtained spectral reflectance of cotton canopy .Besides, the corresponding canopy characteristic information were measured at the same time. The major canopy characteristic information include LAI, MLA, TCDP, TCRP, K, MLD, AFM and ADM. Excel software and SPSS (Statistical Package for Social)are used to process data. Multifactor statistics technique and curve fitness analysis are adopted to discuss the correlation between cotton canopy characteristic information and red-edge parameters as well asits different transformations, which include its first derivation spectral reflectance, red-edge ,hyperspectral characteristic variables, vegetation indices. And then the hyperspectral remote sensing estimation models for different canopy characteristic information are constructed based on the results of correlation analysis, finally, the predictive precisions are analyzed for some chosen models in order to determine the best estimation models for every canopy characteristic information.The correlation analysis of canopy characteristic information and red edge parameters were present. The results indicated: the Depthi, Areai, NDi and EGFN could be used to estimate LAI and AFM; SRo for LAI, MLA and K; dλnir for LAI, MLD and ADM; Lwidth for LAI and AFM . Thereafter, the regression models of retrieving canopy characteristic information, based on red edge parameters obtainable from remotely sensing data, were also established respectively. Red edge parameters could be considered as a sensitive indicator for cotton nutrition status by non-destructino means on a large scale.We analyse the regular pattern of red-edge along with the different growth stages. The results showed that the significantly correlation between NDVI and cotton canopy cover (r=0.6303**,n=62). So we can using NDVI to estimate cotton canopy cover. The slope of red-edge quadratic model has the highest correlation coefficient (r=0.6983**,n=62). It is feasible for using RMSE of estracting cotton canopy cover (RMSE=0.0476g/m2). The study indicates that it can play a vital role in probiding time-specific and time-critical information by using hyperspectral remote sensing for precision farming.For cotton at different growth stages, conducted statistical correlation analysis between various canopy characteristic information and 43spectra parameters, detected out the optimal spectral parameters to make a sound foundation for set up models by using spectral parameters, including statistical correlation analysis on plant MLA,TCDP,TCRP,K,MLD,LAI,AFM,ADM and canopy spectral reflectance spectra. Set up statistical regression model for various cotton canopy characteristic information ,based on spectral parameters.