Quality factor Q estimation and full-space Q field construction

Abstract

Quality factor Q is a fundamental physical parameter for subsurface media, quantifying the absorption attenuation of seismic waves. Accurate Q estimation and full-space Q field construction are essential for enhancing seismic imaging resolution. A complete workflow for Q estimation and full-space Q field construction is developed. First, a near-surface Q estimation method using single-uphole survey data is presented. The effects of inconsistent seismic wavelets and variable geophone coupling on Q estimation are suppressed through multiple spectral ratios. Second, a medium-deep Q estimation method using common midpoint (CMP) gathers is developed. The method exploits attenuation differences among traces with different offsets and avoids explicit wavelet estimation, normal moveout (NMO) correction, and along-layer traveltime picking. Third, the estimated Q values are converted into absorption coefficients, which are fitted and then converted back to Q values to reduce local fluctuations caused by noise and other interference. Fourth, a prediction framework based on long short-term memory (LSTM) neural networks is proposed to reconstruct the missing interval between the near-surface and medium-deep Q fields. Validation with known Q values demonstrates that absorption coefficients provide more stable predictions than layer Q and equivalent Q, while the proposed LSTM strategy achieves higher prediction accuracy than conventional spline interpolation. Field seismic tests demonstrate the reliability of the proposed workflow for near-surface and medium-deep Q estimation and full-space Q field construction. After inverse-Q compensation using the constructed Q field, reflection continuity and weak deep reflections are improved, and the frequency within the analyzed window increases by 13 Hz.