Canopy Spectral Response Characteristic And Yield Quality Model Construction Of Maize Under Drought Stress

Maize is one of the largest grain crops in China,and it plays a very important role in ensuring food security and alleviating the contradiction between grain supply and demand.With the deterioration of the global climate and lack of water resources,the growth and development of maize has been hindered,seriously affecting the yield and quality of maize,and increasing the risk of food security in China.Hyperspectral remote sensing technology can quickly obtain real-time predictions of the growth and nutritional status of crops in the field,as well as the yield and quality,and at the same time diagnose whether the crop has been subjected to adversity stress,providing scientific basis and technical support for the development of smart agriculture.This study is to take the semi-compact maize Zhenghong-505 as the research object,relying on maize-soybean strip relay intercropping system,and set up four different drought stress treatments in the jointing stage,tasselling stage and filling stage of the maize,soil water gradient is 75%± 5%of normal stress(NS),60%± 5%of light stress(LS),45%± 5%of moderate stress(MS)and 30%±5%of severe stress(SS).Measure the canopy spectral reflectance and agronomic parameters(morphological characteristics,photosynthetic pigments,carbon and nitrogen parameters,etc.)of each stage and the yield and quality at maturity,and comprehensively spectral techniques,physiological and biochemical tests and mathematical statistical analysis and other technical means.On the basis of determining the agronomic relationship between agronomic parameters and yield quality at maturity,the best agronomic parameters in each stage are selected,and the spectral data(spectral area parameters,vegetation index and wavelet analysis)of different growth stages and sensitive agronomic parameters and yield quality are analyzed for the quantitative relationship,using linear non-linear regression method and partial least squares(PLS)regression analysis to establish a spectral estimation model of agronomic parameters and yield quality,for maize growth status,physiology.The rapid monitoring of biochemical metabolism and yield quality and field water management provide theoretical basis and technical support.The leaf area index,leaf water content,biomass,photosynthetic pigment content,and leaf carbon and nitrogen accumulation of maize treated with drought stress at different growth stages will gradually decrease,of which the total chlorophyll content,carotenoid content,leaf carbon accumulation,and leaf nitrogen accumulation content decreased the most during the jointing stage,which were 89.64%,94.70%,83.28%and 82.93%,respectively;the soluble sugar content gradually increased with the increase of the degree of drought stress,and the maximum increase under severe stress during the tasselling stage was 100.04%.The maximum increase in jointing stage and filling stage was 58.85%and 31.85%;under severe drought conditions in the tasselling stage,maize yield and starch content decreased the most,34.92%and 36.16%,respectively.Has the largest impact,with a decrease of 23.19%.The crude protein content of the grain will increase significantly under drought treatment,and the maximum increase in severe drought during the tasselling stage is 41.64%.Under drought stress,the canopy spectral emittance of maize will increase in the visible region,1484～1757 nm and 1969～2209 nm,and decrease in the near-infrared region;as the degree of drought stress intensifies,the position of the red edge of the first derivative spectrum The short-wave direction shifts are 725 nm,724 nm,724 nm,and 720 nm,and the amplitude of the red edge gradually decreases.Through quantitative analysis of the agronomic parameters and yield quality of maize at the jointing stage,tasselling stage,filling stage and full growth stage,the agronomic parameters with the best correlation between the yield and quality of each stage and full growth stage and the mature stage were selected.At the jointing stage Chl-b content and total chlorophyll content,leaf carbon content and leaf carbon accumulation during the tasselling stage,leaf carbon accumulation and soluble sugar during the filling stage,and leaf area index and leaf nitrogen accumulation during the full growth stage;to lay the foundation for realizing indirect estimation of output quality by spectral data.Correlation analysis was carried out on the spectral area parameters,the vegetation index of previous studies,the random combination of the bands of the preferred waveband,the vegetation index and the wavelet function,and the selected agronomic parameters.Except for the leaf carbon content and the wavelet function db4694,1 have the best correlation(r=0.606,p<0.01),other agronomic parameters have the best correlation with the vegetation index of random combination of bands,and the correlation coefficient is generally above 0.8,and they all reach a very significant correlation.Using the spectral parameters with the best correlation to establish linear non-linear estimation models of agronomic parameters in different growth stages,the chl-b and total chlorophyll contents at the jointing stage were tested with the quadratic function established by RVI1635,1511 and RVID506,D754,respectively;the carbon content and biomass of the leaves during the tasselling stage were tested by the quadratic function established by db4649,1 and RVID1596,D2012,respectively;and the leaf carbon accumulation was tested by the linear equation established by RVID1596,D2012.The carbon accumulation and soluble sugar content of leaves were tested by the quadratic function established by RVID558,D402 and NDVID779,D589,respectively;the leaf area index of the whole growth stage was tested by the quadratic function established by RVID743,D734.The accumulation of nitrogen was tested by the linear equation established by RVI555,537.The fitting precision of the PLS regression model of maize agronomic parameters based on wavelet function is obviously higher than the linear model and quadratic function model of spectral parameters,but the PLS regression model contains a large number of spectral bands,which makes the model more complicated.According to the technical path of "spectral data-agronomic parameters—yield quality",the spectral monitoring model of maize agronomic parameters is integrated into the agronomic models of agronomic parameters and yield quality,and a maize yield and quality prediction model based on the spectral parameters of the sensitive stage is established.The test accuracy of the quadratic function model of yield and quality and PLS regression model based on the spectral parameters at each growth stage have their own advantages;the comparison found that the ratio of vegetation index RVID558,D402 during the tasselling stage can accurately predict the yield and starch content of maize,and the filling stage.The wavelet function db4691,1 can accurately predict crude protein content,and the ratio vegetation index RVID1596,D2012 can accurately predict crude fat content.Therefore,the use of spectroscopy technology can quickly and non-destructively predict the yield and quality of maize at maturity.