Study On The Growth And Physiological Response To Drought Stress In Winter Wheat Based On Hyperspectral Monitoring

Wheat is one of the three food crops in our country.In recent years,drought and the shortage of agricultural water resources have seriously affected the stability of crop production and food security.As a rapid and nondestructive monitoring technology,hyperspectral remote sensing technology can obtain the plant canopy information and has been widely used in crop production.Timely and accurately monitoring the growth and physiological parameters to learn the crops growth and the water status is significant in help the managers to timely adjust the irrigation schedule and make assessment of crop drought.In this study,the change trends of growth and physiological parameters of winter wheat under different irrigation treatment and the response of each parameter to the canopy spectrum were studied.The relationships between the growth and plysiological parameters and spectral characteristic parameters,spectral vegetation indices,spectrum with different transformation methods and sensitive bands were analyzed,and the accuracy and stability of the models based on different modeling approaches were compared.In addition,the differences between the indices on different scale(blade or canopy)were compared in the monitoring of growth and physiological parameters.The main conclusions for this thesis were as follows:1.The results showed that the treatment had negative effects on the the growth and physiological parameters.Meanwhile,the responses of canopy reflectance and the derivate reflectance to the agronomy were studied.It indicated that with the increase of the growth and physiological parameters canopy reflectance decreased in the visible range,while it would increase in the near-infrared region.For detivative reflectance of canopy spectra,the red edge region was most obvious.2.It indicated that there had close relationships between the parameters of winter wheat and the hyperspectral parameters or the absorption character parameters of continuum-removed spectrum.Among these spectral parameters,there had significant negative correlations between growth parameters of winter wheat and spectral parameters(the reflectance of green peak and red valley(Rg and Rr),the position of green peak(?g),the skewness of red valley(SKr)and the ratio of the reflectance of green peak in red valley Rg/Rr.At the meanwhile,the growth parameters had significant positive correlations with the red edge parameters(the amplitude and area of red edge(Dr and SDr)and the ratio of amplitude and area(Dr/SDr)and the absorption character parameters of continuum-removed spectrum(the maximum of absorption peak(BDmax),the total area of absorption peak(ABD)and right area of absorption peak(RA))in 400-550 nm and 550-770 nm.It indicated that the spectral can be used in the estimation of growth indices in winter wheat.The optimal spectral parameters for each index were selected by the performance of the calibration and validation models.The optimal spectral parameters for leaf area index(LAI),leaf chlorophyll destiny(CHD),leaf nitrogen accumulation(LNA)are RA(550-770 nm),Rg/Rr and BDmax(400-550 nm),respectively.And Dr/SDr is both the optimal spectral parameters for the leaf water content(LWC)and plant water content(PWC).3.DVI(435,447)is the optimal spectral index for LAI estimation.And the optimal spectral indices for the estimations of canopy water content(CWC),LWC and PWC are the derivative spectral indices RVI(1712,1605)?FDRVI(687,531)and FDDVI(688,532),respectively.The optimal spectral indices for chlorophyll concentration(Chl)and CHD are DVI(798,804)and FDDVI(671,802),respectively.The optimal spectral indices for leaf nitrogen content(LNC)and LNA are DVI(1214,1297)and FDRVI(702,688),respectively.4.The results indicted that the pretreatment transformation methods can significantly increase the correlations of transformed spectrum with each indicator.The effects of different pretreatments on the accuracy of PLSR models are different.The spectrum with derivative transformation can significantly improve the accuracy of the calibration model and reduce the number of latent variables(LVs)of the PLSR model,but the performance of the validation is inconsistent.The best pretreatment for each indicator were analyzed by comparing the performance of the calibration and validation models.The optimal PLSR models for LAI was based on the normalized spectrum.While the derivative of the log(1/R)was the best pretreatment for leaf LWC and Chl.The optimal PLSR models for CHD and PWC were based on the spectrum with the transformation of derivative of 1/R and multiple scattering correction(MSC),respectively.And the rest parameters would have a better performance with derivative spectrum.In addition,the results showed that the selection of sensitive bands is influenced by the pretreatment method.The performance of the models with the sensitive bands selected from different transformed spectrum were different.5.Among the different approaches to estimate the models,the PLSR model based on the optimal transformation method is better with high accuracy,low RMSE and stable performance.6.The results showed that the relationship between canopy spectrum and winter wheat growth and physiological parameters on canopy scale were closer.And the estimation models of different approaches performed better in the monitoring of the indicators on canopy scale.