| The disease of the pine trees caused by Bursaphelenchus xylophilus is a destructive disaster. It has the characteristics of fast spreading, great harm and difficulty to control, so as to the enormous losses of forestry and national economy. Therefore, great more importance has been attached to the forecast of this disease. As a new technique, the remote sensing is widely applied in forestry investigation and disease monitory because of the advantages of its highly efficient, convenient and objective. As the remote sensing technology become more mature, the key of development of remote sensing application technology has depended on the innovative methods and the improvement of the equipment.The subject of this study is the disease of Pinus massoniana infected by B.xylophilus which is spotted in Sequoia town, Xiaqu village and Zhukou town that are all located in Taining county, Fujian province. Transpiration rate and spectral characteristics of P.massoniana infected by B.xylophilus in four different disease periods (HP, EPD, MPD, and TPD) were measured to find the changing law; transpiration rate of HP and EPD were then compared with collected spectral data of the same periods and spectral characteristics (REP, RES, RBP, RRP, RERVI, RENDVI) using correlation analysis.9 remote sensing images were selected from Landsat-8 and ZY-3 satellites that were photographed at the same period, and finished their corresponding pre-progressing respectively. One of panchromatic images from ZY-3 was then selected as a reference to register accurately other images. The area of P.massoniana in the study area was extracted from OLI on Landsat-8 satellite, which produced the vector diagram through the classification of Support Vector Machine (SVM) based on the forest resources inventory and planning data in Taining county. The fused images from ZY-3 were then tailored by the former vector diagram of P.massoniana. Finally, a regression model was established to reflect the correlative relationship between the Ratio Vegetation Index (RVI) which is characteristic in the plant’s parameters and transpiration rates of P.massoniana in different disease periods, which would preferably achieve the retrieval of disease grades of B.xylophilus by adopting classification method of the decision tree based on expert experience. The result shows that the maximal value 0.86, which is the correlation coefficient between the raw spectral and the transpiration rate of P.massoniana between HP and EPD, was appeared at the 753nm wavelength. R2 is bigger than 0.7 after fitting analysis between the transpiration rate and red edge parameters, and the result of fitting analysis with RERVI almost up to 0.8515. The accuracy of classification finished through OLI image reaches 87.15%, which met the study criteria. It is currently suitable in the disease condition of study region that the retrieval of disease grades extracted from the fused images from ZY-3 satellite. The conclusions of this study are as follow. The transpiration rate of P.massoniana gradually descends as the disease becomes severer, which can be detected before the damage is visible. Therefore the transpiration rate of P.massoniana can be an effective factor to diagnose the disease early. The transpiration rate has a highly correlation with six spectral characteristic parameters. Among them, the red edge ratio vegetation index (RERVI) shows an extremely significant correlation. Therefore, it is advantageous for the early forecasts of forest disease through the remote sensing technique based on RERVI. Besides, the transpiration rate also has an appropriate correlation with the ratio vegetation index (RVI) of samples in different periods calculated by the images from ZY-3 satellite. As a result, it is possible to achieve the retrieval of disease grades by building a reasonable model through the analysis of ground-air data, equipped with the remote sensing technology. |
PDF Full Text Request