Inversion Of Aboveground Biomass Of Tropical Rain Forest Considering Vertical Structure Characteristics

Forests are the main carbon storage pool of terrestrial biosphere,and forest biomass(AGB)is a key parameter characterizing the carbon sequestration capacity of forest systems.With global climate change and environmental issues becoming increasingly prominent,the ecological services provided by forest systems such as carbon storage have become a research hotspot.In order to address the issues of underestimation and neglect of forest vertical structure information in forest AGB,this article uses multi baseline airborne SAR data obtained by Afri SAR in the Mondah and Lope forest regions of Gabon,Africa as the data source,and uses the Mondah forest experimental area to construct an AGB inversion model,which is validated in the Lope forest experimental area.The main research content is as follows:(1)Research on vertical structure reconstruction methods for tropical forests.In view of the complexity and diversity of the vertical structure of tropical forests and the difficulty in obtaining vertical direction features,this paper studies the reconstruction methods of forest vertical structure in the Mondah and Lope study areas based on the multi baseline heavy orbit Pol In SAR data set,and uses beamforming algorithm Beamforming,multiple signal classification MUSIC algorithm,Capon algorithm and compressed sensing L2,1 norm algorithm to conduct 3D imaging research on echo signals under different polarization modes.Research has found that using Capon algorithm can better describe the distribution of backscattering power of forest scatterers in the vertical dimension,preserve vertical structural information such as canopy and surface,and effectively obtain vertical structural profiles of forest vertical structure and surface information.(2)Method for Extracting Vertical Structure Information of Tropical Forests.Based on the conclusion obtained from(1),the SKP-AS ground scattering decomposition method is used to separate surface scattering and volume scattering.The Capon method is used to reconstruct the vertical structure of the forest dominated by volume scattering and surface scattering,and to extract parameters such as multi-layer backscatter power in the vertical direction,tree crown height,canopy height,understory terrain,and the sum of vertical backscatter power of the forest canopy.The research results indicate that the correlation between the vertical backscatter value and forest AGB is relatively high at 0.784;In addition,the correlation between tree crown height and parameters such as under canopy height,under forest terrain,and the sum of vertical backscatter power of forest canopy is higher at 0.667.(3)Construction of a forest aboveground biomass inversion model.A tropical forest AGB inversion model was constructed based on the vertical structure characteristic parameters of the Mondah forest,and its generalizability was verified using the Lope forest remote validation method.Select forest vertical structure parameters with high correlation coefficients,and use cross validation stepwise regression and support vector regression methods to construct a Mondah forest AGB inversion model,which is compared and analyzed with the forest AGB data measured by Lidar.Subsequently,the applicability of this method was verified using the Lope study area of the tropical rainforest.The research results indicate that the stepwise regression model has the best performance in inverting forest AGB,with a correlation of 0.927 and a determination coefficient of 0.859 in the sample area.The support vector regression method may be more suitable for areas with high biomass of 300-400 tons/ha,with a correlation of up to 0.699.The addition of forest vertical structural parameters will to some extent improve the estimation of forest AGB,especially in tropical forests with complex vertical structural characteristics.