Study On Moisture Monitoring Of Flue-cured Tobacco Based On Hyperspectral Remote Sensing

The objective of this paper is to study the different genotypes flue-cured tobacco leaf and plant water content and spectral characteristics under different water contents(drought stress,waterlogged stress)by using ASD spectrometer.Respectively analyzed the Correlation of fuel moisture content with leaves spectral reflectance and its first derivative,the Correlation of leaf equivalent water thickness with leaves spectral reflectance and its first derivative,and the Correlation of plant water content with leaves spectral reflectance and its first derivative.And then screening the best sensitive band for moisture status of flue-cured tobacco.And estimating models of the flue-cured tobacco water contents were set up by means of the best sensitive band,the spectral reflectance and its first derivative.And the corresponding experiment was done in order to testify the mathematical model.For the use of hyperspectral remote sensing technology for monitoring drought disaster and waterlogging disaster and the development of digital tobacco provide theoretical guidance.The main results are as follows:1.Changes of water content and spectral characteristics of flue cured tobacco under different water treatments.Under different treatments,the changes of the canopy and leaf spectral curves of flue-cured tobacco were consistent,which had the characteristics of green healthy plants.But at different growth stages of different water treatment,the reflectance of each band showed some differences.The performance of different genotypes of flue-cured tobacco was consistent.In different growth period,leaf fuel moisture content,leaf equivalent water thickness and plant water content are characterized by the decrease trend since the degree of drought stress,and the rise trend since the waterlogging degree.For the three water index,the water treatment of each value from high to low in turn for severely waterlogged,moderate waterlogged,normal water demand,mild drought,moderate drought,severe drought,different genotype flue-cured tobacco varieties were consistent.2.Sensitive bands of moisture content in Flue-cured Tobacco.The tobacco leaf and canopy spectral data acquisition,and correlation analysis with three water indicators,screened the best sensitive bands of fuel moisture content t in flue-cured tobacco leaves respectively is located at 2240 nm leaf spectral reflectance(r=0.686,P=0.000)and is located at 408 nm leaf of the first derivative spectra(r= 0.654,P=0.000).The best sensitive bands of equivalent water thickness respectively at 1924-1928 nm,1934-1940 nm,1942-1944 nm of the original leaf spectral reflectance(r=0.700,P=0.000)and is located at 1063 nm leaf of the first derivative spectrum(r= 0.715,P=0.000).The best moisture sensitive bands in flue-cured tobacco plants were located at 1411 nm of the original canopy the spectral reflectance(r= 0.479,P=0.000)and 1103 nm in the canopy of first derivative spectra(r= 0.495,P=0.000).3.Optimal prediction model of moisture content in Flue-cured Tobacco.Used by screening the best sensitive wave bands,and 350-2500 nm band within the scope of the whole spectrum data and its first derivative,respectively,of leaf fuel moisture content,leaf equivalent water thickness and plant water content to build a linear model,multiple regression model and BP neural network model,and with the test data in 2015 as sample to test in the constructed model.We can get that a derivative is more suitable for the inversion of flue-cured tobacco moisture index content than the original spectral reflectance.And,in the construction of the flue-cured tobacco FMC,tobacco LEWT,flue-cured tobacco PWC monitoring model,with the inversion precision optimal BP neural network model,which shows the higher universality and stability,the best effect of monitoring water status of flue cured tobacco.