Estimation Of Forest Above-Ground Biomass Based On GF-3 PolSAR Data And Landsat-8 OLI Data

Remote sensing technology is widely used in the estimation of Regional forest aboveground biomass because of its macro,fast,dynamic and repeatable characteristics,and is the main method of forest aboveground biomass estimation at present.As the first C-band synthetic aperture radar(SAR)satellite of China,gaofen-3(GF-3)can obtain a wealth of spectral information reflected by ground objects,classify different types of forests and extract vegetation index.However,the penetration of C-band SAR is not enough in dense forest,and it is greatly influenced by soil conditions in sparse forest.The combination and optimization of SAR parameters by means of polarization decomposition can improve the estimation of forest aboveground biomass by backscatter coefficient of C-band SAR data,and the complementary information of different remote sensing data sources can be used to accurately retrieve forest aboveground biomass quantity.The research area is Gaofeng planted forest in Nanning City,Guangxi Province.We extracted backscattering coefficient and polarization decomposition parameters from GF-3 PolSAR data and then extracted spectral information,vegetation index,texture from Landsat-8 OLI data.Based on single source remote sensing data and multi-source remote sensing data,we estimated forest aboveground biomass and explores the feasibility of multi-source remote sensing to estimate forest aboveground biomass.In order to effectively use high-dimensional remote sensing features to estimate forest aboveground biomass,this paper proposes a k-nearest neighbor method based on sequence forward feature selection(KNN-SFS).Based on the principle of minimizing the RMSE calculated by the predicted and measured values of forest aboveground biomass,the KNN-SFS method takes the forest sample survey data as a reference,and selects the optimal feature combination successively and iteratively,so as to optimize the KNN estimation model of forest aboveground biomass.As the experimental results show,the accuracy of estimating forest aboveground biomass by combining GF-3 PolSAR data with Landsat-8 OLI data is RMSE=21.05 t·hm-2,R2=0.75,which is better than that by using GF-3 PolSAR data(RMSE=28.38 t·hm-2,R2=0.47)and by using Landsat 8 OLI data(RMSE=29.52 t·hm-2,R2=0.42).To some extent,union of multi-source data can make full use of the advantages of different sensors,optimize the combination of features,reduce the number of features needed to estimate forest aboveground biomass,and reduce the redundancy of simultaneous interpreting.The KNN-SFS method can efficiently and quickly select relevant features from high-dimensional remote sensing features to optimize forest aboveground biomass modeling,which provides a feasible method for forest aboveground biomass estimation.