基于高精度时频瞬时相位谱的多尺度曲率及其应用

Multiscale curvature via high-precision time-frequency instantaneous phase spectrum and its application

基于地震信号瞬时相位估算的曲率属性具有较高的分辨率,被广泛应用于地震数据的不连续性检测。为进一步提高曲率属性的估算精度,提高地震数据对地质构造的刻画能力,基于时频分析方法具有多尺度分辨能力的特点,提出了基于改进的短时傅里叶变换获得高精度时频瞬时相位谱的多尺度曲率属性估算方法。改进的短时傅里叶变换对待分析实信号的解析信号加上对称时窗函数并进行傅里叶变换,从而获得高精度时频瞬时振幅谱、高精度时频初始相位谱和高精度时频瞬时相位谱。为减少随机噪声对高精度时频瞬时相位谱的影响,利用高精度时频瞬时振幅谱的峰值振幅及其对应的高精度时频瞬时相位谱重构输入信号,并对重构信号进行改进的短时傅里叶变换,获得稳定的高精度时频瞬时相位谱,在此基础上进行高精度多尺度曲率估算。利用合成数据及实际三维地震数据的应用结果验证了方法的有效性。

Curvature estimation based on the instantaneous phase of seismic signal has high precision and is widely used for discontinuity detection.To further improve the precision of curvature estimation and the ability of seismic data to describe geological structures,according to themultiscale resolution of tim-frequency analysis,this paper proposes the modified short-time Fourier transform (STFT) to obtain high-precision time-frequency instantaneous phase spectrum for multiscale curvature estimation.The modified STFT applies a symmetric function windowed Fourier transform to the analytic signal of a real signal in order to obtain the high-precision time-frequency instantaneous amplitude spectrum, high-precision time-frequency initial phase spectrum and high-precision timfrequency instantaneous phase spectrum.To minimize the effett of random noise on the high-precision time-frequency instantaneous phase spectrum,the input real signal is reconstructed by the peak high-precision tim-frequency instantaneous amplitude spectrum and its corresponding high-precision time-frequency instantaneous phase spectrum thereafter, the modified STFT is applied to the reconstructed real signal to obtain stable high-precision time-frequency instantaneous phase spectrum for high-precision multiscale curvature estimation.Tests conducted on both synthetic data and a real 3D seismic data verified the effectiveness of the proposed method.