Monitoring Dead Pinewoods In Pine Wood Nematode Disease Area Based On Satellite-UAV-Ground Remote Sensing Data

Pine is one of the plant species for afforestation in China,and it is also the main building timber with high ornamental value and economic value.Since the first discovery of black pine infected with pine wood nematode in Nanjing in 1982,the disease has spread and spread in China's pine forest,causing a large number of pine trees to die,not only destroying the forest ecosystem but also bringing huge economic losses.Therefore,the timely and efficient prevention and control of pine wood nematode disease is the key to curbing the spread of the epidemic.Epidemic monitoring is the basis of the prevention and control of pine wood nematode disease.Based on the Satellite-UAV-Ground Remote Sensing Data,this paper discusses the monitoring method of dead pine trees in the pine wood nematode disease area.The monitoring method mainly uses satellite,UAV multi-source remote sensing image and deep learning method combined with GIS technology to complete the monitoring and positioning tasks of dead pine trees,mainly carrying out the following research contents:(1)The spectral curve data of healthy pine needles and different degrees of dead pine needles were measured by ground non-imaging hyperspectral measuring instrument,and the spectral characteristics and sensitive bands of GF-2 remote sensing images were analyzed.Moreover,the spectral indices suitable for distinguishing between healthy pine and dead pine were selected from the existing 28 spectral indices.(2)Research on the use of the resource 3 satellite stereo image to extract the fine DEM(Digital Elevation Model)data,combined with the VECA terrain correction model for terrain correction of the GF-2 satellite remote sensing image.Using the topographically corrected GF-2 remote sensing image combined with the normalized red vegetation index(RGNDI)obtained from the screening,the distribution of dead pine trees was extracted,and the data of dead pine trees was superimposed on the forestry small-span data to calculate the occurrence of dead pine trees in the unit area.And to create a disaster map of the affected area.(3)According to the severity of the disaster in the study area,the UAV is used for the aerial photography operation in the severely affected areas,and the unmanned aerial photography is used to carry out the identification and positioning of the dead pine trees based on the deep learning method.The method uses the UAV image to make the data set of the deep learning method,and builds and trains the convolutional neural network model with reference to the classic Alex Net model.Finally,the trained convolutional neural network model is used to locate and identify the dead pine trees in different target areas.The actual ground survey sample and the manually labeled result are the true values,and the recognition accuracy of the dead pine tree in different scenarios is evaluated.The results show that the deep recognition method can detect dead pine trees with the highest recognition accuracy of 80%.This method has certain advantages in the monitoring of dead pine trees in UAV images,which can improve the efficiency of manual visual interpretation,and can provide reference for the efficient and rapid research on the identification and location of dead pine trees.