Numerical modelling of near-surface seismic scattering by the partitioned domain method

Ivan Sanchez, William M. Agudelo, Daniel O. Trad, Daniel Sierra

Accurate modeling of near-surface seismic scattering (NSS) is crucial for reducing noise and enhancing the quality and interpretability of seismic data. This process enables the identification of specific characteristics of NSS , leading to the development of more effective techniques to attenuate it. While various methods having been developed to model NSS from elastic wave forward modeling, they often present constraints in considering complex overburdens with substantial topographic variations or random heterogeneous media. In this work, we present the partitioned domian method (PDM) to address these limitations. The method involves two stages of elastic wave modeling to generate synthetic seismic data with and without NSS. The first stage uses the entire earth model to generate the total wavefield (TW), encompassing all seismic wave types. The second stage employs the near-surface portion of the earth model to yield the near-surface wavefield (NW), inclusive of direct waves, Rayleigh waves, and NSS. A deep wavefield (DW), comprising bodywave reflections and refractions from deeper structures, is then derived by subtracting the NW from the TW. To study the performance of the proposed method, we conduct a numerical example using the SEAM Foothills Phase II model. The results demonstrate that our method can effectively model the NSS, enhancing the understanding of land seismic noise that facilitates the development of new attenuation techniques.