Disaster Degree Retrieval Of Dendrolimus Tabulae Based On Hyperspectral Remote Sensing Technology

Dendrolimus tabulaeformis has caused serious damage to Pinus tabuliformis.According to the statistics,there are 0.12 million hectares of destroyed area every year in Jianping county,Liaoning Province of northeastern China,and the direct economic loss is 3.4 million CNY per year.How to monitor and prevent the pine caterpillar pests effectively,improve the quality of Pinus tabulaeformis forest and reduce economic losses,is an urgent problem to be solved.In recent years,remote sensing technologyespecially hyperspectral remote sensing technology has developed rapidly,which can offer processing method for quantifying diagnosis plant chlorophyll content and water content as well,and make a comprehensive analysis of the degree of damage to the affected plants to improve the degree of disaster monitoring accuracy.In this study,hyperspectral remote sensing technology is applied to the monitoring of damage degree of the pest,and the physiological information and spectral information of Pinus tabulaeformis are extracted to study the coupling relationship with damage degree(leaf loss rate).Based on the field survey data and the hyperspectral data of UAV in Jianping county,Liaoning province,China,this paper gets the sensitive spectral indices NDSI,DSI and RSI of chlorophyll content and water content of Pinus tabulaeformis by a correlation analysis method,and inverts the chlorophyll content and water content in a linear regression method,and establishes the model to invert the leaf loss rate of Pinus tabulaeformis with the multiple linear regression and artificial neural network method.The multiple regression model of the higher leaf loss rate is y=0.799-31.85×dDSI(427,811)-142.01×dDSI(797,482)+0.067xdRSI(774,677)whose accuracy is 92.97%,and the multiple regression model of the low leaf loss rate is y=0.13+2.22xNDSI(808,816)+2.53xNDSI(881,920)whose accuracy is 83.1%.The accuracy of artificial neural network method is 78.98%.The results show that the two models can invert the leaf loss rate of Pinus tabulaeformis and the accuracy of multiple regression inversion model is higher.The application of leaf loss rate inversion model to the UAV hyperspectral image is helpful to the multi-scale and multi-temporal data for remote sensing monitoring of disaster.