Geometry-Guided Fourier-Based Mode Decomposition for Diffraction–Reflection Separation and High-Resolution Seismic Imaging

Diffractions provide favorable illumination conditions and contain valuable information about small-scale subsurface discontinuities, making them important for high-resolution seismic imaging. However, their weak energy is commonly masked by strong reflections, making effective diffraction–reflection separation essential for diffraction-based imaging. In this study, we develop a geometry-guided Fourier-Based Geometric-Mode Decomposition (GMDF) method that exploits the distinct physical and geometric characteristics of reflections and diffractions. Specifically, the proposed method incorporates linear-feature and energy-concentration constraints in the frequency–wavenumber (f-k) domain to characterize the high-energy approximately linear signatures of reflections and reduce inter-mode energy coupling. A penalty term is further introduced into the iterative mode-update process to suppress the leakage of nonlinear diffraction energy into the reflection mode. Synthetic and field-data experiments demonstrate that the proposed method effectively separates diffraction energy from reflections and preserves weak diffraction events. The resulting diffraction images provide clearer delineation of small-scale subsurface discontinuities and improve the imaging of fault-related structures. These results demonstrate the potential of geometry-guided GMDF for high-resolution seismic imaging in marine geophysical exploration.