Conclusions

Q estimation is highly affected when the noise is strong. We have proposed a novel multi-channel Q estimation algorithm based on the commonly used spectral ratio method. In order to deal with the spectral nulls, we formulate the spectral division as an inversion problem and solve it using the shaping regularization method with a smoothness constraint. In order to deal with the instability of the solution when the random noise is very strong, we propose to use an extra spatial smoothness constraint in the shaping regularization framework. The synthetic examples with spatially constant and varying Q demonstrate that the proposed method can obtain very accurate Q estimation results. The real seismic data example with strong noise demonstrates that the proposed method can help obtain stable Q estimation performance in the presence of high noise level.