Study On Monitoring Cotton Infected With Verticillium Wilt Based On Multi-platform Remote Sensing

[Objective] This thesis elucidated hyper-spectrum and satellite image characteristics of cotton infected Verticillium wilt, definituded influence factors of monitoring cotton infected Verticillium wilt by remote sensing, established estimating models of SL (severity level) on cotton infected Verticillium wilt, realized accurate reorganization of cotton infected Verticillium wilt using remote sensing technology, and provided reference for monitoring cotton other diseases and insects.[Method] The data which from near ground hyper-spectrum, satellite image and field research were analysized using original reflectance,first derivative spectrum,spectral continuum removal method and anova analysis in 2005-2009 years. The characteristics of spectrum,the change of agronomy parameters and the influence factors of monitoring about cotton infected Verticillium wilt were ascertained. Meanwhile, the estimating models of SL were established on cotton infected Verticillium wilt on the base of relatation analysis and regression analysis, and the accuration of estimation models were assessed using RMSE(Root Mean Squar Error),RE (Relative Error) and r (relatation coeffectance) etc. In addition, the images were preprocessed and analyzed by ENVI 4.3 software in experiment areas. Using the supervised classification on the base of parallelepiped, the cotton fields of suspected Verticillium wilt were classed. Finally, the results of classes were evaluated by overall accuracy,user's accuracy, producer's accuracy and kappa coefficient[Result] Through monitoring cotton infected Verticillium wilt using multi-platform remote sensing, this paper got progress as follows:(1) The spectral characteristic of cottons infected Verticillium wilt was ascertainedCompare with normal cotton leaf, spectral reflectance of leaf infected Verticillium wilt was significantly increased in visible light region (400-700 nm) and near-infrared region (700-1 300 nm) with the increase of the SL. Compare with spectrum reflectance of single leaf infected Verticillium wilt, spectrum reflectance of canopy infected Verticillium wilt were similar in visible light red region (620-700 nm), but it inverted in near-infrared region (700-1300 nm) with the increase of the SL of disease. The red edge swing on cotton canopy and single leaf infected Verticillium wilt all decreased, red edge position all equal moved to the blue. The absorption band position, the absorption depth and the absorption area all decreased, however the absorption width all increased for cotton canopy and single leaf infected Verticillium wilt. In addition, with the increased of the SL of single leaf F0 increased, Fm, Fv, Fv/Fm, Fv/F0, Fm/FO,ΦPSII, ETR and qP all decreased, but NPQ were first up and then down in the chlorophyll fluorescence parameters.The agronomy parameters Chl.a,Ch1.b,Chl.a+b,WC and TN all decreased gradually, T increase little by little, Car was first down and then up, THK was first up and then down with the increased of the SL in cotton leaves infected Verticillium wilt. The agronomy.parameters Ch1.a, Ch1.b, Ch1.a+b, Car, LAI, FB, LNC, WC, TNC, CD and DB all decreased gradually, T increased little by little, SNC and ONC increased in overall trend.Those results indicated there were a better spectral respond on the agronomy parameters of cotton canopy and single leaf infected Verticillium wilt.(2) The estimating models of cottons infected Verticillium wilt were estblished by remote sensing.The thesis indicated that 434-724 nm and 909-1600 nm were selected out as sensitive band region to single leaf SL of Verticillium wilt.1001-1110 nm and 1205-1320 nm were selected out as sensitive band region to canopy SL of Verticillium wilt. The band 698 nm were selected out as characteristic band of single leaf SL of Verticillium wilt, and 806 nm were selected out as characteristic band of canopy SL of Verticillium wilt. Those quadratic and linear models in which had variable (R 700 nm/R 825 nm) and (R 806 nm-R 1455 nm) could invert accurately the cotton canopy and single leaf infected Verticillium wilt, and they may be used to forecasting the position of cotton infected Verticillium wilt, quantitatively. The linear model in which had variable NDVI[702,758] could invert accurately the pigments contents in cotton single leaf infected Verticillium wilt, and it can be commended to estimating the pigments contents in cotton single leaf infected Verticillium wilt. Those linear models in which had variable image spectral indexes of DVI and DNSIb had best estimated precision. The models of RMSE and RE were lower and the slop and intercept were approach 1 and O.They were commended as best models to estimate SL of disease cotton by satellite image.(3) It was analysized on influence factors of monitoring the cotton infected Verticillium wilt by hyperspectrum and satellite image.The spectral characteristics of cotton leaves infected Verticillium wilt had been not influenced by time of testing, place of testing and varieties of cotton. The spectral characteristics of cotton leaves infected Verticillium wilt had been minor affected by the cotton stresses of Salt injury, Fusarium wilt, Red spider and Aphid. The spectral characteristics of cotton leaves infected Verticillium wilt had been major affected by location of leaves on plant, different parts of a leaf, different faces of a leaf and testing instrument. The spectral characteristics of cotton leaves infected Verticillium wilt had been large affected by the cotton stresses of Drought, Nitrogen scarcity and Angular leaf spot. However, if the testing background was selected improper, it would have inaccurate result. On the other hand, the spectral characteristics of cotton canopy infected Verticillium wilt had been not influenced by time of testing, place of testing and varieties of cotton. The spectral characteristics of cotton canopy infected Verticillium wilt had been minor affected by the cotton stresses of Drought, Fusarium wilt, Red leaf blight, Red spider and Aphid. The spectral characteristics of cotton leaves infected Verticillium wilt had been large affected by height and angle of testing and the cotton stresses of Salt injury, Nitrogen scarcity. In addition, The image characteristics of cotton infected Verticillium wilt had been large affected by the cotton stresses of Red spider, Drought and Salt stress in single phase image,and had been few affected by Red spider, Drought and Salt stress in multi-phase image.(4) The cottons infected Verticillium wilt were monitored by satellite remote sensing.The result which on the base of TM satellite monitoring to cotton infected Verticillium wilt demonstrated that the health and disease cotton fields had obvious difference in the major cotton growth stages, it was feasible to diagnose cotton field of disease through the interpretation symbol of satellite image. TM4 band of satellite could be regarded as the best wave band to monitor cotton field of disease; the peak bell set (from the last ten-day of July to mid Aug) could be regarded as the best period to monitor disease of cotton with TM image. Using the supervised classification on the base of parallelepiped, the cotton fields of suspected Verticillium wilt were divided into three classes. The tested results indicated the overall accuracy and kappa coefficient were all over 85% at the crucial growth periods in two years.