宽频带地震数据瞬时谱分解及快捷解释方法

瞬时谱分解产生的若干共频率瞬时谱数据体不仅占用大量存储资源，而且影响了地震资料解释的效率。为此，本文首先提出联合应用短时傅里叶变换及S变换的时频谱分解方法，提高中、低频段时间分辨率及分解宽频带地震数据；接着应用分频段主成分分析实现宽频带瞬时谱数据集的降维与优化，并保持不同频段信息随频率变化的特性；最后将聚类测度引入模糊C均值聚类中，对经分频段主成分分析后数据集做进一步自适应融合。实际资料应用效果证实，本文方法可有效分解宽频带地震数据，快捷地提取和突出包含在大量瞬时谱数据集中的主要信息，清晰地刻画储层几何形态和展布特征，节省了存储资源并提高了宽频带地震资料解释效率。

Instantaneous spectrum decomposition and fast interpretation of broadband seismic data

Several iso-frequency instantaneous spectral data volumes generated from spectral decomposition of broadband seismic data (BBSD) not only occupy a great deal of storage resource, but also make the interpretation become huge workload. In order to overcome these problems, we propose a new solution. First we introduce a new BBSD spectral decomposition based on Fourier transform and generalized S transform (GST), which can improve time resolution of time-frequency spectra at low frequency. Then frequency band principal components analysis (DFBPCA) is proposed to implement the dimensionality reduction and optimization of broadband instantaneous data sets, which can maintain information characteristics with frequency. Finally a cluster separation measure is introduced into the fuzzy C-means clustering (FCM) to implement further adaptive fusion of data sets from DFBPCA. According to real data application, the proposed method decomposes BBSD into iso-frequency instantaneous spectral data sets with good time-frequency resolution, extracts fast and highlights the primary information contained in a large number of instantaneous spectral data sets, and depicts spatial distribution characteristics of reservoirs while reducing storage resource and improving BBSD interpretation.