Seismic Q estimation with logarithmic spectrum equation root
Cao Siyuan1,2, Tan Jia2, Gao Ming2, Yuan Dian1, Yang Jinhao1, Zhang Haoran1
1. State Key Laboratory of Petroleum Resources and Prospecting, China University of Petroleum (Beijing), Beijing 102249, China;
2. CNPC Key Laboratory of Geophysical Exploration, China University of Petroleum (Beijing), Beijing 102249, China;
3. Institute of Geophysics, Research Institute of Exploration and Development, Xinjiang Oilfield Company, PetroChina, Urumqi, Xinjiang 830013, China
Abstract：Quality factor Q is an important parameter, one of characteristics of subsurface medium absorption of seismic wave. Q describes internal essential characteristics of the media, and is also an indication factor of oil and gas recognition. Therefore the accurate determination of seismic Q has certain significance to reservoir prediction. In general, it is stable to extract Q in the frequency domain, and the commonly used methods include spectral ratio and centroid frequency shift. Although the spectral ratio method has a higher theoretical precision, but it is vulnerable by the influence of SNR and has lack of stability. On the contrary, the centroid frequency shift method has higher robustness. But this method needs to do theoretical approximations, and it is difficult to do error analysis. So we present in this paper an inversion method based on statistical combination of logarithmic spectrum to calculate Q according to natural logarithmic spectrum of the wavelet amplitude attenuation law in the strata. Based on this combination method of statistical properties, we propose logarithmic spectrum equation root method by integral for frequency. According to computations on model data, the combination of statistical properties has higher theoretical precision, and avoids the assumption of source spectrum using in the centroid frequency method; the logarithmic spectrum equation root method combines the advantages of the spectral ratio method and the centroid frequency method; and the accuracy and denoise are much higher than spectral ratio and centroid frequency.
