基于局部均值分解的地震信号时频分解方法

Seismic Data Time-Frequency Decomposition Based on Local Mean Decomposition

在分析地震资料时,因吸收和衰减等原因,地震信号往往呈现出非稳态性。时频分解方法能将地震信号分解成多个稳态的子成分,从而为描述和分析地震信号的属性提供了便利,如短时傅里叶变换、小波分析、经验模式分解(EMD)等方法。本文引入了一种新的时频分析方法——局部均值分解(LMD)。该方法将地震信号按其时频属性分解成多个乘积分量信号(PFs),较EMD分解所得的固有模态函数(IMFs)保留了更多的局部信息,同时模态混叠效应更少。对模型数据和实际数据的处理结果验证了LMD方法能够合理地分解地震信号,更准确地描述地震资料中不同时间尺度的构造信息,为进一步的地震数据处理和解释提供参考。

In seismic data analysis, it is a general case that seismic＇ signal always displays nonstationary characteristics because of wave-attenuation effects. By time-frequency decomposition methods, such as short-time Fourier transform, wavelet transform, and empirical mode decomposition, seismic data can be decomposed into a set of stationary components, which are easier to be processed and interpreted. Following this idea, the paper introduces a new method named local mean decomposition （LMD）. The LMD method can decompose seismic data into several product functions （PFs）. Compared with the intrinsic mode functions （IMFs） by the EMD method, the PFs preserves more details and the mode mixing effect is weaker. The applications to model data and field data show that the LMD method can make the decomposition more properly, and capture the local character of seismic data from different time points.