融合气象因子和多时相Sentinel-1数据的森林蓄积量遥感估测OACHSSCD

To enhance the reliability of forest volume estimation models and the spatio-temporal consistency of forest volume products,aiming at the perturbation problem of meteorological factors on the time-scale stability of polarization information,a Meteorological Disturbance Index(MDI)is proposed to quantitatively evaluate the degree of interference of meteorological factors on SAR(Synthetic Aperture Radar)data.By integrating the MDI index and a stratification strategy,a reconstruction method for multi-temporal remote sensing features is established to improve the remote-sensing inversion accuracy of forest volume from multi-temporal SAR data and the reliability of related products.In this study,30 scenes of Sentinel-1 dual-polarization SAR images in 2022 and the real-time product dataset of the China Meteorological Administration's Land Data Assimilation System(CLDAS-V2.0)are used as the main data sources.First,by analyzing the response relationship between meteorological factors,multi-temporal remote sensing features,and their sensitivity to forest volume,the MDI is put forward,and the interference of meteorological factors on the temporal stability of multi-temporal remote sensing features is quantitatively evaluated.Then,in combination with the forest volume stratification strategy,the perturbation degree of meteorological factors on the polarization characteristics of different forest stand structures and their forest volumes is explored in depth.The MDI and the stratification strategy are integrated to construct a reconstruction method for multi-temporal remote sensing features.On this basis,a forest volume estimation model is constructed using the forward feature selection method and four regression models,and the influence degree of meteorological factors on the forest volume estimation model and its results is further analyzed.The research shows that meteorological factors are an important factor leading to the temporal instability of polarization characteristics extracted from SAR data.The MDI index proposed in this paper can quantitatively evaluate the perturbation degree of meteorological factors on polarization characteristics.Through the feature recombination by integrating MDI and the stratification strategy,the sensitivity between remote sensing features and forest volume can be significantly enhanced.Compared with the multi-temporal averaging method and the feature-level combination method,the model accuracy and reliability of forest volume estimation obtained from the alternative feature set by the reconstruction method proposed in this paper are significantly improved.Among them,the R2 value ranges from 0.39 to 0.54,the rRMSE value ranges from 23.82％to 27.61％,and the forest volume estimation accuracy is improved by 4-6 percentage points.The feature reconstruction method that combines MDI with forest volume stratification can effectively improve the temporal instability of forest volume estimation using SAR images.The recombined features effectively improve the estimation accuracy of forest volume,which has important application value for forest resource monitoring and management.