Estimation Of Forest Aboveground Biomass Based On Remotely Sensed Data Through Considering Background Reflectivity Variations

Accurately estimating the spatio-temporal variations of forest aboveground biomass(AGB)is the prerequisite and foundation of quantitatively simulating the carbon sequestration capacity of a forest ecosystem.It is also helpful to assess the carbon budget and better understand the responses of forest ecosystem to climate changes and human activities.Remote sensing technique has been widely used in retrieving forest canopy leaf area index(LAI)and change detections due to its large spatial coverage and short revisiting abilities.The empirical relationship between the biomass of forest canopy and leaves serves as the theoretical foundation for retrieving forest AGB based on forest canopy LAI using remotely sensed data.However,the complex forest vertical structure consisting of overstory and understory(i.e.,shrubs.grass,tundra,bare earth and snow)introduces errors in estimating forest AGB using optical remotely sensed data.In this study,we first combined the four-scale geometric optical model and Moderate Resolution Imaging Spectroradioemeter(MODIS)data for extracting the reflectivity of forest background,and then produced the forest canopy LAI maps with and without considering the forest background reflectivity variations.By doing this,the effects of forest background on forest AGB were quantitatively assessed.Finally,the national forest inventory(NFI)data were used to cross-compare the forest AGB results estimated from remotely sensed data.Based on our results,we concluded that:(1)Using the MODIS BRDF data and a semi-empirical kernel-based bidirectional reflectivity model,we calculated two reflectivity images at specific view zenith angle(VZA)angles(i.e.,0° and 40°).Combined with the ratio of light canopy,light background,shaded canopy and shaded background simulated from the four-scale model,forest background reflectivity could be obtained.By analyzing the relationship between forest canopy and background reflectivity at both nir and red band,we testified the importance of considering the impacts of background on forest structure parameters inversion.(2)Based on MODIS reflectivity data and the four-scale model,we estimated the LAIe(MODIS LAIe)under two scenarios.In the first scenario,the forest background reflectivity map extracted from the first step was inputted into the LAIe inversion algorithm.While the constant background reflectivity value was inputted into the LAIe algorithm in the second scenario.Then,the LAIe results were validated with high resolution LAIe products estimated from Landsat TM data(TM LAIe).It showed that after using the dynamical background reflectivity values,the R2 between MODIS LAIe and TM LAIe increased from 0.36(n = 25,p<0.05)to 0.49(n = 25,p<0.01)in Tahe county,and from 0.02(n = 28,p>0.5)to 0.12(n = 28,p<0.5)in Genhe county,respectively.(3)By combining MODIS-based LAIe,clumping index(CI)and specific leaf area(SLA)for different forest types,we estimated forest canopy AGB based on the relationship between filed-based leaf biomass and forest canopy AGB.Then the forest AGB results were validated by the ones obtained using NFI-based approach.The results showed that the AGB result with dynamical background reflectivity values had a stronger relationship(R2 = 0.86,n = 10,p<0.01)with NFI-based AGB than that(R2 = 0.52,n = 10,p<0.05)with constant background value.This research developed a generic method to estimate forest aboveground biomass based on remotely sensed data through considering background reflectivity variations assess its effects on the final accuracy of mapping forest AGB at regional level.In addition,the spatio-temporal variations of forest AGB were investigated accordingly,which indicated that the important role that forest background plays in retrieving forest AGB.This work provide a foundation for better understanding the carbon cycle and budge in the temperate forest ecoregion of China.