Background Small-scale discontinuous geobodies such as faults, collapse columns, and pinch-out points are widely present in underground spaces. They are closely associated with the safe production and development of underground resources such as coals, oil, and gas. As the wave field responses of small-scale discontinuous geobodies, diffracted waves can overcome the shortcomings of traditional reflected wave imaging, being capable of accurately identifying and localizing small-scale geobodies.

Objective and Methods  This study aims to achieve diffracted wave imaging of discontinuous geobodies. Based on the differences in kinematic and dynamic characteristics between reflected and diffracted waves, this study leveraged the precise self-adaptive decomposition ability in the time and frequency domains of the variational mode decomposition (VMD) method, as well as the efficient and stable global optimization ability of the grey wolf optimizer (GWO). The VMD-GWO algorithm effectively avoided empirical errors and local optimum problems while possessing elevated accuracy of diffracted wave separation and high self-adaptability.

Results and Conclusions  Compared to the whale optimization algorithm (WOA) and ant colony optimization (ACO), the particle swarm optimization (PSO), sparrow search algorithm (SSA), and GWO yielded lower optimal fitness values (3.172), indicating their higher optimization performance. Moreover, compared to PSO and SSA, GWO demonstrated faster convergence, achieving the global optimum through only six iterations. These results highlight the superiority of GWO in both optimization performance and efficiency. The tests of synthetic and actual data verified that the proposed GWO-VMD algorithm is effective in diffracted wave separation and shows strong suppression of reflected waves, thereby enabling the high-resolution imaging of microscale structures.