谱图重排的谱分解理论及其在储层探测中的应用

Reassigned time-frequency decomposition and its application of hydrocarbon exploration

谱分解理论是把单道地震记录分解为连续的时频谱平面,是地震资料处理和解释的重要技术之一.由于谱分解方法的多解性,所以同一道地震记录因为分解方法不同,得到的时频谱是不一样的.短时傅里叶变换,小波变换,S变换和匹配追踪算法都是对信号开窗分析,这些方法都受到不确定性原理的限制.Wigner-Ville变换避开了不确定性原理的限制,但是交叉项的存在限制了本方法的使用.本文利用谱图重排的时频分析方法(RSPWV)对合成的单道地震记录和实际的地震资料进行了分析.与短时傅里叶变换和匹配追踪算法的比较得出:此方法具有较高的时频分辨率,能够很好地识别气层.

Spectral decomposition refers to any method that produces a continuous time-frequency spectrum of a seismic trace.Time frequency analysis is an important technology in seismic data processing and interpretation.Since spectral decomposition is a non-unique process,thus a single seismic trace can produce various time-frequency spectra.The short time Fourier transform（STFT）,wavelet transform（WT）,S-transform and matching pursuit（MP） method are all using windows,thus under the control of uncertainty principle.The Wigner-Ville distribution（WVD） avoids the windows.However,WVD is quadratic that produces cross-terms for multi-components signal that restricts its applications.In this paper,we analyze the time-frequency properties of a synthetic trace by using the reassigned smooth pseudo Wigner-Ville distribution（RSPWV） method,experiments and field data show that the method is superior to the STFT and MP methods,and can delineate the gas reservoir.