Tropical Forest Biomass Estimation Using Airborne P-band Tomographic SAR

In forest resource inventory and monitoring,forest height,forest volume,forest aboveground biomass(AGB)and understory topography are four important parameters.In these parameters,forest AGB plays an important role in the global ecology and carbon cycle.It is one of the hot issues in forestry remote sensing that how to estimate forest AGB on a global or regional scale accurately.Synthetic aperture radar(SAR)has become one of the main techniques of forest resources monitoring with its advantages of day-and-night and all-weather observation.Compared with the X,C and L-band SAR,P-band SAR has a better penetration performance in forest monitoring,and it has advantages in forest parameter inversion.Tomographic SAR(Tomo SAR)is an advanced SAR technique developed in recent years.Tomo SAR has strong ability of three-dimensional information acquisition and can describe the spatial structure of forest,which is beneficial to the forest AGB estimation.Due to the advantages of P-band SAR and Tomo SAR in forest monitoring,the Tomo SAR data of P-band was chosen to estimate the forest AGB in this study.The study was carried out over the forest of Paracou in French Guiana,and the study data was obtained by the airborne experiments of Tropi SAR 2009.The main research work includes: Tomo SAR imaging and forest vertical profile extraction,the features mining for forest AGB estimation and Tomo SAR forest AGB estimation.The main contents of this paper are as follows:(1)Tomo SAR imaging and forest vertical profile extractionFirstly,the digital terrain model(DTM)and digital surface model(DSM)of the study area were extracted by Pol Tomo SAR,and the DTM was used for terrain phase removal and topographic compensation.After that,the forest profiles obtained by Capon,MUSIC and Beamforming were evaluated by the matching between Tomo SAR profiles and Li DAR.Finally,we selected the Tomo SAR forest vertical profiles obtained by Beamforming to estimate AGB.Through the study,we can get the following conclusions.DTM and DSM with high accuracy can be obtained by Pol Tomo SAR;DTM extracted by Pol Tomo SAR can be used in Tomo SAR data processing,such as terrain phase removal and topographic compensation;Compared with Capon and MUSIC imaging algorithm,Beamforming imaging algorithm is more accurate for describing forest vertical profile,which is suitable for Tomo SAR forest vertical profile imaging.(2)Forest AGB estimation based on Tomo SAR backscattered power distributionFirstly,we fitted the backscattered power curve based on the Tomo SAR forest vertical profile imaged by Beamforming.Then the law between the backscattered power distribution and forest AGB levels was explored by analyzing the morphological changes of the curve with different AGB levels.Finally,we extracted the estimation features based on the backscattered power distribution with different forest AGB levels and the AGB was estimated by extracted features.The experimental analysis showed that the BPC-4 and GVPR-19 proposed in study had good correlation with forest AGB.The results showed that it had the best estimation accuracy with GVPR-19 and BPC-4 as the features.The results of forest AGB estimation with GVPR-19 as single feature were slightly poor,while it had the worst result with BPC-4 as single feature.Through the study,we can get the following conclusions: With the increase of forest AGB level,the main part of backscattered power in the Tomo SAR vertical profile will move up.When the forest AGB level reaches a certain higher level,the main part of backscattered power will be in the position of forest canopy;BPC-4 and GVPR-19 proposed by the law of the backscattered power distribution have good correlations with forest AGB,so they are suitable for forest AGB estimation;The method based on the distribution of backscattered power distribution has better estimation accuracy,which can be used in the estimation of forest AGB.(3)Forest AGB estimation based on Tomo SAR and forest investigation factorsFirstly,the study analyzed the forest investigation factors that affect forest AGB at stand scale and the factors related to Tomo SAR backscattered power were summarized.Then,five forest AGB estimation features,BPV-5、BPV-31、LBPC、BPFAH and FAH,were proposed based on forest investigation factors and Tomo SAR backscattered power distribution,and forest AGB estimation was carried out based on the proposed features.The experimental results showed that BPV-5、BPV-31、LBPC、BPFAH and FAH extracted by this method were well correlated with forest AGB,and the correlation coefficient between BPFAH and AGB was 0.706,which had the best correlation with forest AGB.The results of forest AGB estimation showed that compared with the two existing methods,the accuracy of this method had been improved to a certain extent.Through the study,we can get the following conclusions: The estimation features based on Tomo SAR and forest investigation factors have good correlation with forest AGB,which is suitable for forest AGB estimation;The forest estimation method of Tomo SAR combined with forest investigation factors can better reflect the information of forest vertical profile,and has better accuracy than the existing estimation methods.The innovation of this study can be summarized into three aspects: First,this paper reveals the law of Tomo SAR backscattered power distribution with different forest AGB levels It is found that the main part of Tomo SAR backscattered power in forest vertical profile will move up with the increase in forest AGB level.When the forest AGB level reaches a certain higher level,the main part of Tomo SAR backscattered power will be in the position of forest canopy.Secondly,this paper proposed a forest AGB estimation method based on the backscattered power distribution of Tomo SAR,in which two new forest AGB estimation features are proposed,including BPC-4 and GVPR-19.The accuracy analysis of estimation results shows that the proposed method achieved good estimation accuracy,which shows the effectiveness of the forest AGB estimation method based on Tomo SAR backscattered power distribution.Thirdly,the paper proposed a forest AGB estimation method based on Tomo SAR and forest investigation factors.In this method,in addition to the backscattered power and forest height,we also proposed two new forest AGB estimation features including LBPC and BPFAH.The results show that compared with the existing methods,the method based on Tomo SAR and forest investigation factors has better accuracy,and this method also explores a new idea for the study of Tomo SAR forest AGB estimation.