Monitoring Tobacco Influenced By Mosaic Virus Based On Remote Sensing Technology

Hyperspectral remote sensing technology has been widely used in agriculture in recent years. And agriculture remote sensing has also been involved in all aspects of agricultural production, with its advanced methods and techniques for precision agriculture and efficient agriculture. Tobacco,one of the most important economic crops in China, is an important part of the national income. Whereas, the production of tobacco is deeply affected by mosaic virus(TMV) which is one of the main diseases of tobacco. And this virus brought tremendous economic losses to farmers brought huge losses because the tobacco mosaic virus has a high incidence rate with a field morbidity. To get the information of tobacco infected by mosaic virus on real-time and accurately is of great importance so that the tobacco mosaic virus can be prevented and, if not entirely prevented, to at the very least be timely and efficiently controlled. is of great importance so that the tobacco mosaic virus can be prevented and, if not entirely prevented, to at the very least be timely and efficiently controlled. The study established models for the diagnosis on tobacco mosaic disease and figures of temporal spatial distribution of tobacco mosaic disease through ground-based hyperspectral remote sensing as well as remote sensing image of high resolution so as to achieve rapid non-destructive monitoring of tobacco mosaic virus.The experiment took Yicheng, Daotuo of Yishui County, Shandong Province as the object of the study which was conducted from June to September in 2014. In the study, the leaf and canopy spectral data of different disease grades was obtained from Field Spec4 ASD portable hyperspectral radiometer, the relative value of chlorophyll in tobacco was measured by HSY-051 chlorophyll meter, and the moisture content of tobacco leaves was measured in the lab. Then the disease severity estimation models were established through the analysis of correlation between the spectral data which contains the first-order derivative, vegetation index, various position variables and the area variables and the physiological and biochemical indexes as well as disease grade. Finally, the tobacco mosaic disease severity was checked and inversed qualitatively and quantitatively with the use of ZY-3 satellite images of high resolution. The main results of the study are as follows:(1) The original spectral curves of different disease grades for tobacco were studied. The result shows that the spectral reflectance of the leaves infected by the mosaic virus increases in the visible region(400-700nm) and obviously decreases in the near-infrared region(750-1300nm); through analisizing the first derivation of different disease grades for tobacco, it could be concluded that the red edge moved to the blue light. What's more, with the increase of the degree of disease, the above changes are more and more obvious. Tracking the relative reflectivity of mosaic tobacco in different periods of time we found that the mosaic virus would become severer with the advance of growth stage and the illness would spread more rapidly at the late growth stage.(2) The correlation analysis between disease index and chlorophyll content was carried on. As a result, there is a significant negative correlation between disease index and chlorophyll content, that is, with the increase of disease index, the chlorophyll content decreased. And the relationship between the two could be expressed by y=-42.01x+42.223, among which R2 is up to 0.8675. It is showed that we can indirectly assess the degree of tobacco diseases by estimating the chlorophyll content of tobacco. Then the first order derivative, vegetation index, various position variables and the area variables are extracted from the data of hyper spectrum. And the author adopts the above three parameters combined with stepwise regression analysis to establish chlorophyll estimation model of tobacco influenced by mosaic disease. The results indicate that the chlorophyll estimation model based on position variables and area variables is superior to the other two kinds of estimation models based on first order derivative or vegetation index; The optimal chlorophyll estimation model of tobacco influenced by mosaic disease is: y=109.658+21.208(?)+115.622Rg-2.230(?)-65.138(?)-161.06?r(R2=0.885); the inspection accuracy is 76.86%.(3) The correlation analysis between disease index and moisture content was carried on. It is found that the two time formula y=-31.484x2+9.5045x+81.11 could express the relationship between disease index and water content, and its R2 is 0.712. It is suggested that we can assess the degree of tobacco diseases approximately by estimating the water content of tobacco. Then the correlation analysis of water content with hyperspectral reflectance and its eight transforms as well as three spectral index was proceeded. The result showed that the optimal water estimation model is established based on WBI. That is Y=-2.5236-0.7472X1+1.5293X2-2.8204X3+8.7505X4-5.4376X5+1.7175X6, among which, the value of X1 to X6 are R2298/R2333?R2299/R2333?R2321/R2344?R2322/R234?R2322/R2346?R2322/R2348, and its coefficient is 0.7716. The correlation coefficient between the measured value and the estimated value of the model is 0.8784, and the coefficient of determination is 0.7686.(4)According to the canopy hyper spectral data of mosaic tobacco in different levels which was measured at tobacco fields in Yishui in 2014, the 11 commonly used for estimating the severity of disease were calculated. And the correlation analysis between the above vegetation indexes with disease index was proceeded. The principle of variable importance in projection(VIP) was used to select the best indicator variable for disease condition. Then the partial least squares method was used to establish the disease severity estimation model. At the same time using ENVI5.1 to process the ZY-3 satellite image of high resolution, and using the estimation model and satellite images to monitor the spatial distribution of disease severity. The results showed that: the ratio vegetation index(RVI), difference vegetation index(DVI), and normalized difference vegetation index(RDVI), transformed vegetation index(TVI), soil adjusted vegetation index(SAVI) as indicators of mosaic tobacco could effectively estimate the disease severity. The determination coefficient of the model based on the above 5 index was 0.8165. The model was used in remote sensing image and thus the tobacco mosaic disease grade map was obtained. In order to verify the accuracy of remote sensing monitoring, 140 sampling points in field were chosen for disease investigation and were recorded the geographic coordinates. At the same time, the disease degree of each sampling point was examined in remote sensing image. As a result, the accuracy of remote sensing monitoring was 77.13%. So it could realize the effective monitoring of mosaic tobacco.