笔者通过对多种连续小波基的平稳序列时频的分析，提出了经验模态的非平稳序列的时频分析方法，并在地震资料处理解释应用中作探索性的试验。试验结果表明：当一个地震记录道长为5s长和2ms采样的资料，经验模态分解后可获得6—7个固有模态函数单分量，分解的第一个分量C₁(t)具有较宽的频率成分，可用作地震属性的主分量分析。

 This article in the continual wavelet base's stationary series of time-frequency analysis method foundation, proposed based on the empirical mode decomposition non-stationary series of time-frequency analysis method, and interpretation of seismic data processing applications were exploratory tests. The results show that a seismic trace is 5 seconds long and 2 ms sampling data， obtained 6-7 IMF single components after decomposed, the decomposition of the first component C1(t) with wide frequency components can be used as the principal component analysis of seismic data.