Fusing Multi-temporal Optical And InSAR To Monitor The Damage Degrees Of Shoot Beetle In Yunnan Pine Forest

Forest pests are one of the important threats to the healthy growth of forests,and the monitoring of its damage is of great significance to forest protection.The limited optical data in cloudy rain areas make forest pest monitoring difficult.So a method of monitoring the degree of forest pests by using interferometric synthetic aperture radar(InSAR)and optical data is proposed.Xiangyun County of Yunnan Province was selected as the study area and the multi-temporal C-band Sentinel-1 images and Sentinel-2 image were applied.Based on the information of the time-varying characteristics of SAR parameter and spectral characteristics of optical image,the paper explored the feasibility of using multi-temporal data and multi-source data for the classification of health forest and different degrees of damaged forest.The results showed that:(1)In the single image of SAR,the difference between backscattering coefficient and coherence coefficient of healthy forest and damaged forest is very small,so it is necessary to integrate the temporal information and interference information and fuse the multi-temporal InSAR data for classification.(2)The time-varying characteristics of coherence coefficient and backscattering coefficient are analyzed by combining the phenological phase of Yunnan pine and relative humidity in the height of 2 meters.The temporal variation of the backscattering coefficient and the coherence coefficient are related to the phenological phenology of Yunnan pine.The correlation between the relative humidity and backscattering coefficient is higher than coherence coefficient,which reached to 0.78 in the mildly damaged forest.(3)Through the field data validation,classification accuracy of the multi-temporal coherence coefficient is higher than the backscattering coefficient,the descending images has the highest precision which reached to 83%.The result showed that the coherence coefficient of C-band SAR time series can effectively identify whether the forest is healthy or with different degrees damage.(4)The fusion of multi-temporal InSAR data and optical data can improve the classification accuracy,up to 89.26%.Sentinel-2 is rich in spectral information and has advantages in distinguishing healthy and mildly damaged forests.Multi-temporal InSAR data has better recognition effects on healthy forests and heavily-damaged forests.The fusing of multi-source data could complement each other and improve classification accuracy.The method has certain advantages in monitoring and classification of forest pests in cloudy areas,as well as to further enhance the ability of remote sensing to monitor pests.