Reservoir Attenuation Characterization Based on Adaptive Frequency Slice Wavelet Transform

Zhang, Yan (Research Institute of Petroleum Exploration & Development) | Zheng, XiaoDong (Research Institute of Petroleum Exploration & Development) | Lu, Jiaotong (Sinopec Geophysical Corporation)

OnePetro 

Summary

Frequency slice wavelet transform (FSWT), a new timefrequency signal analysis method proposed by Zhonghong Yan et al. (2009), has some new properties compared to continuous wavelet transform (CWT). FSWT has been successfully used to process and analyze mechanical vibration signals. In order to test this new transform’s advantage, we try to study reservoir attenuation characterization by FSWT time frequency analysis. In the first step, we will calculate the center of gravity in the seismic signal frequency domain (CGF) and then recognize CGF parameter as a discriminant criterion for FSWT aimed to control the frequency resolution ratio of the seismic signal, which means we could apply the self-adaptive FSWT to seismic signal. After that, the arguments of gradient and intercept will be obtained from the point spectrum in the FSWT time-frequency domain, which is available to research the reservoir characterization. At the end, the Marmousi II model and real data will be utilized to test and verify the effectiveness of the algorithm.

Introduction

Fourier transform could provide global information of frequency of stationary signal but is incapable of supplying local frequency variation about the non-stationary signal. Seismic signal has the unsteady and nonlinear characteristics which are expressed by the statistical variation in space and time domain and Many methods have been popularly adopted to analyze seismic signal including short-time Fourier transform (STFT), CWT, Wigner Ville Distribution (WVD), S-transform (ST), Hilbert-Huang Transform (HHT), Local time-frequency decomposition (LTFD) and Matching pursuit (MP). All these spectral decomposition methods could convert seismic signal from one dimensional time domain to two dimensional time frequency domain, which provide more information and illustrate more features which couldn’t be found in time domain representation. Nevertheless, these methods have some limitations in the practice. Take STFT as an example, as long as the window function is determined the time-frequency resolution will be limited. WVD has the well-known drawbacks including cross-term interference and aliasing, which contaminates the achieved high resolution. The MP method contains several parameters and is a relatively expensive and timeconsuming method.