A Local Time-varying Q-Factor Estimation Method Based on Multidimensional Nonstationary Smoothing

Abstract The quality factor (Q-factor) reflects the attenuation characteristics of seismic waves within geological formations and is widely used to analyze the absorption and scattering properties of subsurface media. Conventional Q-factor estimation methods such as those based on the local time frequency transform (LTFT) and local centroid frequency shift (LCFS) are fundamentally limited by fixed smoothing radii, which inadequately address the nonstationarity of seismic data across time, frequency, and spatial dimensions. To overcome this limitation, this study introduces a nonstationary, locally time-varying Q-estimation approach based on an adaptive smoothing radius selection strategy. By incorporating a variable smoothing radius, our method adaptively models spatiotemporal information, resulting in significantly enhanced stability and noise robustness for Q-factor estimation under complex and dynamic conditions. Validation using both synthetic and field seismic data demonstrates that the proposed method can accurately resolve Q-factor variations, even under severe noise conditions (SNR < 0 dB). Additionally, owing to its flexible framework, the proposed approach exhibits strong scalability and can be extended to a broader range of time-frequency spectrum analyses and Q estimation techniques.