改进型W变换方法及其应用

A time frequency method of seismic wave based on W Transform

低频信息是地震勘探和储层表征的重要依据。由于传统时频分析方法难以描述低频区域的时频特征,因而严重影响了储层表征的精度。W变换方法通过构建包含时变主频的窗函数,可以清晰地刻画低频区域的时频分布,但是,面对高时间分辨率的储层表征需求,单参数控制的W变换不能很好地调整时频分辨率。为此,提出了改进型W变换(MWT)的时频分析方法,可以更为灵活地调整时频分辨率。首先,设计了一个新的斜率参数,结合原有尺度因子,得到包含双参数组合的窗函数。接着,替换原W变换的窗函数获得时间域的改进方法,并将改进型W变换由时间域转换到频率域,实现简单、高效时频变换。与W变换的比较表明,改进型W变换方法利用斜率参数和尺度因子可以更加精细地控制时频分辨率。单子波和子波组合的合成记录实验证明了该方法可以灵活地调整时频分辨率。实际工区数据应用结果表明,改进型W变换方法通过选取合适的参数集,可以在油井储层区域对应出现较为明显的低频异常现象,可以更好地用于储层表征。

Low-frequency information is commonly used in seismic exploration and reservoir descriptions.However,the conventional time-frequency analysis method is too wide in the low-frequency area window and adopts a constant window change function at different times,which makes it difficult to precisely describe the time-frequency characteristics of the low-frequency area.Thus,the accuracy of the description of the reservoir characteristics is significantly affected.By building a window function containing the time-varying dominant frequency,the W-transform method can clearly describe the time-frequency distribution in the low-frequency area.However,in the face of the requirement of high temporal resolution,the W transform controlled by a single parameter cannot flexibly adjust the time-frequency resolution.Therefore,in this study,we propose a novel time-frequency analysis method based on the W transform,called the improved W transform(MWT),to flexibly adjust the time-frequency resolution.First,a new slope parameter was designed,which was combined with the scaling factor in the original W transform to form a new window function.Subsequently,this new window function was introduced into the original W transform to form the MWT method in the time domain.The time domain of the MWT was then transformed into the frequency domain to obtain a simple and efficient time-frequency distribution.Compared with the W transform,the proposed method achieves a flexible time-frequency resolution in low-frequency regions through the joint action of the slope and scale factors.The synthetic experiments of single wavelets and multi-wavelets demonstrate that the proposed method has a more desirable time-frequency resolution than the other methods.The time-frequency distributions in the real seismic data further demonstrate that the proposed method is reasonable and effective and can well characterize the reservoir in the low-frequency area.