The Study To Monitor The Situation Of Cotton Drought Stress By Infrared Thermal Imaging And Hyperspectral Remote Sensing

[Research object]:This study focused on the different water conditions in key growth stages ,between the information of cotton canopy spectral reflectance characteristics, Crop Water Stress Index and canopy characteristics.The purpose of drought stress is a factor to cotton as a measure of the situation of water deficit benchmark, and the adoption of quantitative hyperspectral remote sensing of water stress on cotton timely monitoring of the situation for future large-scale application of hyperspectral remote sensing quantitative monitoring of cotton to provide the theoretical basis of drought.[Research methods]:The use of different water treatments in the cotton-like Honda synchronous data acquisition of hyperspectral remote sensing, canopy temperature data and agronomic data, and spectral features extracted parameters, coefficient of drought stress (CWSI) and Leaf area index(LAI), canopy coverage(Cg), Aerial parts fresh biomass(AFM),Aerial parts dry biomass(ADM),seed cotton fiber yield(kg·hm^-2) and quality indicators, such as agriculture, analysis between these three aspects of the correlation between CWSI in the degree of water stress as a measure of the scale, based on hyperspectral remote sensing of cotton drought monitoring regression model.[Research results]: (1) The study through species water testing and comparison tests, through the Wet Artificial Reference Surface (WARS) extracted with the use of natural CWSI reference extracted from the wet surface of CWSI to do analysis of variance, results showed that both methods extracted CWSI no significant differences, therefore, can make use of man-made wetland reference surface and a combination of infrared thermal imaging to extract the CWSI, and can make use of cotton in Xinjiang CWSI drought stress conditions in real-time monitoring. CWSI with cotton LAI and canopy coverage analysis and found that full bud period of CWSI and canopy cover to the best linear correlation (r=0.9913^**, RMSE=0.0847); full bell a view to CWSI and power index LAI for the best (r=0.9872^**, RMSE=0.2361). cotton CWSI and seed cotton yield and fiber quality indicators related to the main analysis, results showed that the CWSI Micronaire with the most relevant, both showed significant correlation between the logarithm (r=0.7348^**, n=120); seed cotton yield and CWSI was also highly significant correlation between the logarithm (r=0.6894 ^**, n=120); cashmere long (mm) and strength (cN/tex), respectively, with the CWSI was highly significant linear correlation.(2) Through the cotton of LAI, canopy coverage of remote sensing data and statistical correlation analysis, results showed that, LAI and Cg and canopy spectral reflectance correlation coefficients were high platform area in the 740nm-1340nm; parameters, through the establishment of the spectral filter LAI and Cg estimates and model performance for the full bloom were the highest coefficient of determination, of which, NDVI [1210,857] and LAI showed significant correlation between the index, NDWI [1240,860] and LAI showed a significant quadratic function relationship; Rg/Rr and Cg were significant linear correlation between (r_Rg/Rr=0.8820^**, n=120), P_Depth [560] and Cg was significant correlation between the quadratic function (rP_Depth[560]=0.8839^**, n=120).(3) Through the cotton CWSI and canopy reflectance spectroscopy and statistical correlation analysis showed that the 730nm-1350nm wavelength spectral reflectance and CWSI showed significant negative correlation to form a trough, and the correlation coefficient reaches the maximum at 780nm department; cotton CWSI and spectral analysis of the characteristics of the relevant variables and found that based on the Rg/Rr of CWSI high precision linear regression model (r_{full bloom}=0.9290^**, n=32), (Rg-Rr)/(Rg+Rr) to be better all models as a quadratic function (r_{full bloom}=0.9376^**, n=32); cotton CWSI and reflection peaks characteristic parameters of absorption valley correlation analysis and found that P_Area[560] with the CWSI in the full bud and full bloom stage the establishment of are the regression model fitting equation one dollar three times the correlation coefficient, the higher accuracy (rfull bud=0.9310^**, r_{full bloom}= 0.9379^**, n=24), Depth[980] with the CWSI from the bloom and bell to the opening of bolls to establish the regression model fitting equation are indicators of the highest correlation coefficient; CWSI in cotton associated with the vegetation index analysis, the selected normalized difference vegetation index NDVI[560,670] and NDVI[980,890] as the best monitoring variables, based on the Normalized Difference Vegetation Index and CWSI cotton the best single-phase spectral model to estimate the main equation to one dollar three times the highest degree of fit.