不依赖震源的归一化全局能量互相关全波形反演

Full waveform inversion of source-independent normalized global energy cross-correlation

全波形反演利用叠前地震数据的走时、振幅和相位等全波形信息反演地下介质参数。但常因地震信号低频成分缺失、初始模型不准确而产生周期跳跃现象导致反演陷入局部极值,同时震源子波和振幅信息的准确度也严重影响着反演质量。笔者提出了一种不依赖震源的归一化全局能量互相关全波形反演方法,可有效结合褶积波场、归一化波场与能量波场的特点从而提高全波形反演的稳健性。数值试验表明该方法可以缓解振幅失真对反演的影响,并在震源子波不准且缺失低频情况下构建可靠的大尺度速度结构,结合不依赖震源互相关全波形反演方法可以构建高精度的地下速度模型,且抗噪性较好。

Full waveform inversion(FWI)retrieves subsurface velocity parameters using waveform information such as travel-time,amplitude and phase of prestack seismic data.However,due to the absence of low-frequency components of seismic signals and the inaccuracy of initial model,the inversion often falls into local extreme values due to periodic jump phenomenon.Meanwhile,the accuracy of seismic wavelet and amplitude information also seriously affects the quality of inversion.A new full waveform inversion method of source-independent normalized global energy cross-correlation is proposed,which can effectively combine the characteristics of convolution wave field,normalized wave field and energy wave field to improve the stability and applicability of full waveform inversion.Numerical experiments show that the proposed method can alleviate the influence of amplitude distortion on inversion and construct an accurate large-scale velocity structure under the condition of inaccurate wavelet and absence of low-frequency of the source.Combined with the full waveform inversion method independent of the source cross-correlation,an accurate underground velocity model can be constructed,and the proposed method has good anti-noise performance.