Multicomponent seismic data reconstruction via quaternion matrix factorization algorithm
Similar to the traditional single-component P wave seismic exploration and acquisition,multi-component seismic data acquisition also encounters missing traces and the irregular sampling problems,and also need to do seismic data regularization.However,for the reconstruction of multi-component seismic data,the component-wise reconstruction methods are currently commonly used,where each component is treated independently.These scalar approaches ignore the nonlinear relationship between the components and destroy the vector properties of the seismic wave field.This paper combines quaternion theory with matrix factorization algorithm and proposes a method for multi-component seismic data vector reconstruction based on quaternion matrix factorization algorithm,and applies it to the vector joint reconstruction of 3C-3D irregular missing data.Compared with the existing component-wise reconstruction methods,this proposed method can achieve joint reconstruction of 3C data,and effectively maintain the nonlinear orthogonal structure properties between different components during reconstruction and protect the vector characteristics of the wave field.Moreover,this method is also an SVD-free algorithm with low computational cost.Finally,synthetic data and field data experiments demonstrate the effectiveness and advancement of the presented method.
Multicomponent seismic dataSeismic data vectorial reconstructionQuaternion matrix factorizationLow rank matrix
万方数据
维普
