三角形网格有限元法逆时偏移的面法向量导向滤波方法OA

The finite element method(FEM)employs triangular/tetrahedral mesh discretization,offering inherent advantages in fitting rugged topography and complex structural boundaries.While reverse time migration(RTM)achieves high-precision imaging of intricate geological bodies,its results are frequently contaminated by low-frequency noise,thereby degrading profile resolution.Integrating the geometric adaptability of the FEM with the imaging fidelity of RTM enhances reliability in complex regions;however,the imaging filtering stage suffers from spatial filtering deviations induced by irregular mesh neighborhoods,rendering it difficult to directly transfer filtering operators developed under the finite difference method framework.To address this issue,this study proposes a face normals-guided filtering approach tailored for triangular meshes.The method first applies bilateral filtering to the face normals of individual elements,preserving normal features at the element centroids.Subsequently,the filtered normals information is propagated to mesh nodes via area-weighted averaging of adjacent triangles to update the imaging amplitudes.During wavefield extrapolation,spatial discretization employs an optimized matrix bandwidth technique to reduce computational overhead,while temporal integration uses the Newmark scheme to ensure numerical stability.Tests on field seismic data from the Tahe area demonstrate that the proposed method achieves a Pratt's figure of merit(FOM)improvement of 6.51%compared to conventional mean filtering.Model experiments further confirm that this approach effectively suppresses RTM low-frequency noise while distinctly preserving stratigraphic boundary characteristics,thereby providing a viable technical pathway for enhancing seismic imaging in rugged topography and structurally complex zones.