| Forest above-ground biomass(AGB)is an important basis for the evaluation of forest carbon stocks.With the increased requirements for the measurement of forest carbon fluxes and carbon stocks from carbon neutral research in recent years,higher requirements have been put forward for the effective and accurate measurement of forest AGB over large areas.Remote sensing tools have great potential for forest observation due to their large observation range and low time and labour costs.Forest height is an important parameter of forest resources and is correlated with forest AGB.Several effective methods have been developed for the estimation of forest height and forest AGB to achieve effective observations.Satellite-borne synthetic aperture radar(SAR)has become an increasingly common remote sensing tool for forest AGB observation tasks because of its all-weather,all-day observation capability and its freedom from cloud interference and estimation saturation from optical sensors.The purpose of this paper is to address the errors caused by temporal decoherence in the inversion of forest height by spaceborne SAR and the low accuracy in estimating forest AGB.Based on the three-stage inversion method and the RVoG model,an improved inversion method is developed for inversion of forest height based on spaceborne PolInSAR,and the simulation accuracy of forest height reaches the best RMSE of 2.3125m and~2 of 0.8126.Subsequently,based on the obtained inversion forest height,a regression model is developed to estimate forest AGB by combining other PolSAR extracted parameters.The purpose of this paper is to discuss the simulation capability of the spaceborne PolInSAR on forest AGB and the sensitivity of the inverse forest height on forest AGB.The highest simulation accuracy of RMSE reaches 1.309 t/ha and~2reaches 0.7516,and the inverse results of forest height obviously increase the accuracy of forest AGB estimation.Summarizing the above results,the following conclusions were obtained in this paper:1.Satellite polarised interferometric SAR(PolInSAR)data has a longer time baseline and greater temporal decoherence interference than airborne data.In addition,the satellite data has a much lower vertical wave number due to the much higher sensor height from the ground than the airborne data.The smaller vertical wave number further amplifies the temporal decoherence interference when extracting forest heights,resulting in significant inversion errors.2.By modelling the effects of phase and coherence due to temporal decoherence,this paper uses a portion of the field measurements as a control,introduces a complex parameter and a real parameter based on the RVoG model,and performs parameter optimisation by triple iteration to extract the optimal parameter values,which are brought into the improved RVoG model,ultimately extracting forest height with an accuracy higher than 15%.3.The sensitivity of forest height to forest AGB was analyzed in a model for estimating forest AGB from spaceborne SAR data.It was found that using only the backscatter coefficients and polarization decomposition parameters extracted from PolSAR data,either using multiple linear regression models or support vector machine regression methods,the forest height data were effective in improving the robustness and accuracy of the PolSAR parameters for estimating forest AGB. |
