Forest Canopy Height Inversion Research Based On Spaceborne Polinsar

Pol InSAR can not only be applied to detecting physical structure properties due to multi-channel polarimetric basis,but also can be applied to measuring location and spatial distribution based on conventional SAR interferometry.Especially for the detection of complicated scene with multiple scattering mechanisms,PolInSAR can effectively separate interferometry phase center via scattering matrix decomposition technique inside resolution cells,and extract specific polarimetric interferometry phase.Consequently,Pol InSAR is of great potential in distinguishing and extracting parameters from objectives in remote sensing,and widely used in research such as extraction of surface parameters,estimating aboveground biomass and forest dynamic monitoring.Inversing forest canopy height is a hot topic in PolInSAR.The data in this paper is TerraSAR-X/TanDEM-X quad-polarimetric and ALOS PALSAR FBD provided by DLR and JAXA.Based on double-bounce and volume scattering model—RVoG and S-RVoG?Sloped Random Volume over Ground?model,we test the behavior of Three Stage Algorithm in forest canopy height inversion,and the validation of results is verified by field survey.The main context is as follows:?1?Forest canopy height inversion based on ALOS PALSAR dual-polarimetric data.Applying SAR image of HH and HV,this paper chooses Three Stage Algorithm based on RVoG model to inverse canopy height.The experimental result is full of invalid data,due to the influence of temporal decoherence and errors in co-registration.?2?Forest canopy height inversion based on RVoG model with TerraSAR-X/TanDEM-X quad-polarimetric data.This paper adopts strategy which combines RVoG model with Three Stage Algorithm,and validates the result comparing to field survey.Research shows that,correlation coefficient,ME and R_MSE of rubber plantation is 0.449,-0.02 m and 4.52 m,while that of natural forest is 0.198,-3.45 m and 6.92 m.Rubber plantation has a better inversion accuracy.In terms of slope in rubber plantation area,the inversion accuracy is highest in level I?0—5°?.With growing terrain slope,inversion accuracy decreases and has a lowest accuracy in level IV?26°—35°?.?3?Forest canopy height inversion based on S-RVoG model with TerraSAR-X/TanDEM-X quad-polarimetric data.Conventional RVoG model ignores forest underlying DEM,causing inversion accuracy reduced by terrain slope.This paper introduces slope corrected S-RVoG model,and adopts strategy using this model with Three Stage algorithm.Research shows that,correlation coefficient,ME and R_MSESE of rubber plantation is 0.527,0.46 m and 3.82 m,while that of natural forest is 0.409,-2.36 m and 5.30 m.Natural forest has a more obvious increasing accuracy.In terms of slope in rubber plantation area,inversion results indicate S-RVoG model improves accuracy of forest canopy height inversion to some extent,and compensates errors caused by terrain slope.According to experiment above,this paper shows the strategy applying RVoG model with Three Stage Algorithm to inverse forest canopy height is effective.Furthermore,improved S-RVoG model compensates errors caused by forest underlying DEM,and improves the accuracy of forest canopy height inversion.The ideal to apply terrain slope compensation provides reference for other PolInSAR forest height inversion algorithm.