Extended centroid frequency shift method for near-surface Q estimation

Seismic waves are often strongly attenuated in the near-surface. As a measure of attenuation, the quality factor (Q) is usually estimated by frequency-based methods. However, most common approaches are limited or affected by the various assumptions, frequency band selection, and noise. We propose an extended centroid frequency shift method without Q value assumption and strong wavelet adaptability. At the same time, through synthetic experiments, we found the reason for the inaccuracy of the traditional centroid frequency shift method-the inadaptability of the skewness that varies with the frequency band. The new approach perfectly overcomes this weakness, making the estimation more accurate and robust. Finally, this method can select spectra with a high signal-to-noise ratio due to its frequency band insensitivity, allowing better noise immunity.