Spectral Element-Based Seismic Modeling of the Complex Near-Surface with Meter-Scale Heterogeneities.

Abstract Seismic exploration in arid regions presents significant challenges due to intense scattering noise generated by highly heterogeneous near-surface layers. These complex geological conditions necessitate the development of advanced processing algorithms capable of accurately characterizing wave propagation phenomena. To systematically investigate near-surface effects, we have developed an efficient and precise forward modeling tool based on the Spectral Element Method (SEM), specifically designed to simulate seismic wave propagation in strongly heterogeneous media. Our implementation incorporates several innovative computational strategies to achieve both accuracy and efficiency. The method capitalizes on SEM's high-order approximation capabilities and its flexibility with unstructured grids, enabling faithful representation of small-scale, high-contrast near-surface inclusions. The computational framework features adaptive mesh refinement, maintaining fine discretization in shallow heterogeneous zones while employing coarser elements in deeper, more uniform strata. Furthermore, we have developed an optimized time-stepping algorithm that automatically determines stable time increments while minimizing computational overhead. This integrated approach results in a robust modeling toolkit that accurately simulates seismic wave propagation through complex near-surface environments characterized by strong heterogeneity and impedance contrasts. The methodology provides an essential foundation for developing and testing advanced processing algorithms aimed at mitigating scattering noise in arid region seismic data.