Study On Forest Height Inversion Using PolInSAR Data

With the gradual deepening of the application of remote sensing technology in the field of forestry,synthetic aperture radar has achieved many important research results in the investigation of forest resources with its unique advantages.Polarimetric interferometry SAR integrates the advantages of Polarimetric SAR and interferometry SAR,and with its full development in theory and application,it has gradually become an indispensable technology in the field of forestry,which forest tree height inversion has become one of the most successful applications of Polarimetric interferometry SAR technology.Forests are the largest terrestrial ecosystems and the basis of the global carbon cycle.Forest tree height inversion is important for understanding the entire ecosystem and is one of the important contents of global change research.Polarimetric interferometry SAR has the advantages of all-day,all-weather,penetrability,high resolution,and low cost.It provides an ideal remote sensing method for the height inversion of large areas of forest vegetation,and has important practical application value.Based on the theoretical knowledge of Polarimetric interferometry SAR,this paper takes BioSAR2007 airborne SAR data as the data source,and takes forest density and time decorrelation as the entry point to carry out the research on the forest height inversion of polarization interference SAR.The main research contents of the paper are as follows:(1)The traditional polarimetric interferometry SAR forest tree height inversion algorithm is analyzed,including DEM difference method,complex coherent amplitude method,three-stage algorithm,coherent phase and amplitude joint inversion method,and the above algorithm is analyzed and verified by using simulated data and real data.(2)The influence of forest density on the traditional algorithm is analyzed by using simulation data,and the coherent phase and amplitude joint inversion method is improved according to its influence,then the simulated data and real data were used to analyze and compare it with the traditional algorithm.Knowing the forest density,this method has certain advantages,with R2=0.92 and RMSE=1.82.(3)On the basis of the introduction and analysis of three kinds of forest tree height inversion models with time decorrelation,the RMoG model is improved by ignoring the ground movement of vegetation in order to solve the problems of initial value dependence,long time and unstable inversion results in RMoG model inversion.The improved RMoG model avoids the nonlinear optimization process when solving the improved RMoG model by reducing the unknowns.Finally,the real data is used to verify the improved RMoG model.The results show that the improved RMoG model shortens the inversion time but does not reduce the inversion accuracy,which proves the feasibility of ignoring the ground motion of vegetation under a short time baseline.Among them,the RMoG model R2=0.47,RMSE=4.17,improved RMoG model R2=0.53,RMSE=6.24.(4)Based on the simulation data,the influence of time decorrelation caused by the change of soil moisture on the inversion result of forest height is analyzed.The result shows that the deviation caused by the change of soil moisture is small,which is far less than the error of the algorithm itself,so the change of soil moisture can be ignored when analyzing the error of the algorithm...