Method for near-surface three-dimensional velocity modeling based on first-arrival

Abstract

In seismic exploration, three-dimensional near-surface tomography is key to solving complex statics problem and is an important prerequisite for migration imaging. To this end, this paper starts with the principles of ray tracing theory and tomographic inversion methods. Based on a comprehensive comparative analysis of the advantages and disadvantages of traditional ray tracing methods and numerical algorithms for tomographic inversion, using the first arrival times in three-dimensional seismic data, a high-precision three-dimensional ray tracing method based on multi-retracing technology and a tomographic inversion method constrained by prior information such as small refraction and micro logging throughout the process are proposed. First, this paper use near-surface survey information such as micro logging and small refraction to constrain the establishment of an initial velocity model and constrain tomographic inversion to improve the vertical inversion accuracy of the model. Secondly, to address the loss of shallow layer information due to excessive shot-to-detector distances, this paper introduces a virtual detector point technique. By adding one or more virtual detector points between known shot points and detector points within a close offset range, the density of rays at near offsets is increased, thereby improving the lateral tomographic imaging accuracy of the near-surface. At the same time, various constraints are introduced during the inversion process, including the range of velocity restriction, inversion slowness correction size constraints, internal iteration number constraints of tomographic inversion, dual-grid inversion, velocity extrapolating and smoothing, etc., greatly improving the accuracy of inversion and the quality of tomographic imaging. Given the large computational volume of three-dimensional data and the high memory consumption of large sparse matrices, this paper employs compressed storage and tomographic matrix solving techniques, greatly saving memory space and enhancing computational efficiency. The test results on theoretical models and actual data show that the ray tracing method used in this paper provides essentially correct ray propagation paths, consistent with geological laws. The final inversion results effectively reveal the velocity and thickness of near-surface low-speed layers and deceleration layers, demonstrating the correctness and effectiveness of the inversion method proposed in this paper. Keywords: Velocity modeling, Ray tracing, Tomographic inversion, Micro logging, Small refraction