Estimation of the Near-surface Q Distribution from Pre-stack Surface Seismic Data

Abstract

Knowledge of the near-surface distribution can be helpful in surface seismic data processing to compensate for shallow absorption effects likely to damage our seismic images. In this paper, a tomographic scheme is proposed to derive 3D near-surface distributions from pre-stack surface seismic data. The approach is based on the inversion of the attenuated travel times of the first arrivals (diving waves); where the attenuated travel time of a propagated wavelet is defined as the integration of both velocity and effects along its propagation path. Such seismic attributes are computed from log spectral ratio linear regressions of the first arrivals and decomposed into distributions by a simple linear inversion. This attenuated travel time tomography is integrated into a first-arrival tomography scheme for the derivation of both a near-surface velocity model (travel time inversion) and a near-surface model (attenuated travel time inversion). The approach is illustrated here using synthetic and field data.Knowledge of the near-surface Q distribution can be helpful in surface seismic data processing to compensate for shallow absorption effects likely to damage our seismic images. In this paper, a tomographic scheme is proposed to derive 3D near-surface Q distributions from pre-stack surface seismic data. The approach is based on the inversion of the attenuated travel times of the first arrivals (diving waves); where the attenuated travel time of a propagated wavelet is defined as the integration of both velocity and Q effects along its propagation path. Such seismic attributes are computed from log spectral ratio linear regressions of the first arrivals and decomposed into Q distributions by a simple linear inversion. This attenuated travel time tomography is integrated into a first-arrival tomography scheme for the derivation of both a near-surface velocity model (travel time inversion) and a near-surface Q model (attenuated travel time inversion). The approach is illustrated here using synthetic and field data.