利用小波阈值法提高地震勘探信号的信噪比

Using wavelet threshold function to improve signal SNR for seismic exploration

在地震勘探信号处理中,能否得到高信噪比的信号是正确解读实际勘探信号的关键。在实际勘探中,受到环境等因素的影响,检波器接收到的信号中夹杂着各种相干及随机噪声,降低接收信号的信噪比。鉴于传统傅里叶变换无法处理随机噪声,本工作提出利用改进的小波阈值法提高含随机噪声的信噪比,通过MATLAB仿真和勘探实验验证。建立仿真模型,比较了改进阈值函数与现有软、硬阈值法的去噪效果;信噪比和均方根误差结果显示,改进阈值函数去噪后的信噪比优于软和硬阈值函数的信噪比。利用地震雷克子波建立了地震正演模型,对比改进阈值函数处理前后的结果显示,改进阈值函数法能充分保留有用信号,有效地消减勘探信号中的随机噪声,可为勘探信号解释提供一种新的去噪方法。

In signal processing of seismic exploration, it is critical that one can acquire high-SNR （signal-to- noise） signal in order to interpret the actual exploration data with accuracy. In fact, many uncertainties, due to environmental factors in the actual explorations, may often cause signals received by the detector to contain various coherent and random noises and thus greatly decrease the signal＇s SNR. To overcome the limitation of traditional method of using Fourier transform in eliminating random noise, this work proposed an improved wavelet threshold method and subsequently verified its de-noising performance through simulations and actual applications, using SNR and MSE （mean square error） as the evaluation indexes in the simulations. The results show that （1） a comparison of the statistics of SNR and MSE using different methods indicates the improved threshold function is superior to the soft and hard threshold functions in de-noising performance; and （2） when Ricker wave was used in the construction of the seismic forward model, it can be concluded, by comparing the results before and after de-noising, that the improved wavelet threshold method can retain the useful signal and effectively eliminate random noise in the exploration signal. It is believed, therefore, that the improved wavelet threshold method can serve as a new de-noising method in the interpretation of exploration signal.