Analysis And Research On Vertical Structure Parameters Of Multi-baseline Tomographic SAR Forest

Forests are the largest terrestrial ecosystems in the world and an important part of the global carbon cycle.Information on forest vertical structure parameters is of great significance for estimating forest aboveground biomass(AGB).Multi-baseline tomographic SAR is an emerging cutting-edge technology,which can detect the spatial structure information of forests in the vertical direction,and has been successfully applied in forest parameter estimation,urban 3D reconstruction and other fields.However,spaceborne SAR still faces severe problems and challenges in 3D tomography because it is difficult to obtain data with many samples and a uniform baseline.In this paper,the spaceborne ALOS-2 L band data is used as the research data,and the Shanghai memorial forest in the Sai Hanba forest farm is used as the research object.The multibaseline tomography technology based on spaceborne SAR data is systematically and deeply studied.An improved multi-baseline tomographic SAR imaging method based on DEM for phase correction is proposed.Using five polarization modes combined with Beamforming and Capon multi-baseline tomographic SAR imaging methods,the results are in good agreement with the corrected LIDAR DEM and DSM.A multi-baseline tomographic SAR forest vertical structural profile.According to the acquired tomographic images,the understory topography and the forest height were inverted,and a model of forest AGB was established by tomographic backscattering.The results show that the Beamforming method of HH-VV polarization obtains the highest accuracy of forest terrain inversion(Bias =-0.11 m,Std = 3.84 m,RMSE = 3.84 m).The highest accuracy of forest height inversion(Bias = 2.19 m,Std = 3.43 m,RMSE = 4.07m)was obtained by the Capon method of HH polarization after the height gain of the scattering phase center of the canopy along the altitude direction was 4.5m.The average difference between the scattering phase center of the canopy and the top of the canopy is 4.78 m,and the ratio to the average height of the forest is 24.91%,this value can provide reference for the underestimation of forest height inversion by SAR.The forest AGB estimation with RMSE = 20.29tons/ha,P=25.83 % was obtained by modeling the forest AGB using the Beamforming method under HH polarization at the height of 14 m with Tomo SAR backscattering.The verification results show that the improved multi-baseline tomographic SAR algorithm can significantly improve the ability to invert the vertical structure parameters of the forest under the condition of uneven baseline and under-sampling of spaceborne SAR data.This study provides a new idea for the use of multibaseline SAR data for tomography in the case of inhomogeneous baselines and data undersampling,which is helpful to realize super-resolution height-dimensional imaging of spaceborne SAR data.