一种引入吸收衰减的压缩感知薄储层识别方法

A Thin Reservoir Identification Method Based on Compressed Sensing with Introducing Absorption and Attenuation

常规基于压缩感知稀疏反演方法是基于频域平稳地震褶积模型进行的,而实际地下介质是黏弹性的,这使得该方法的反演反射系数振幅缺失、位置偏移。针对该问题,通过引入时频衰减因子,构建改进的衰减感知矩阵,将稀疏反演从常规的频率域拓展至衰减频率域。反演结果在一定程度上可恢复缺失的振幅,增强弱信号的识别能力。由于噪声影响,以上处理结果中仍存在噪声干扰,故在衰减频率域稀疏反演的基础上,引入平滑的高斯函数,对反演目标函数进行优化以压制残存的噪声干扰;之后将反演结果与褶积宽频子波可生成高分辨率地震剖面,由此形成了一种引入吸收衰减的压缩感知薄储层识别方法。薄层理论模型及含有河道砂储层的实际地震资料的处理结果表明,本文方法较常规方法,有效地增强了薄层弱信号的振幅及横向连续性,可在保证信噪比的情况下,提高地震资料的分辨率及薄储层的识别能力。

The conventional sparse inversion method of reflection coefficient based on compressive sensing is carried out on the frequency domain stationary seismic convolution model,but the actual underground medium is viscoelastic,which distorts the amplitude and position of the inversion reflection coefficient.To solve this problem,an improved“attenuation sensing matrix”was constructed by introducing the time-frequency attenuation factor,which expanded the sparse inversion from the conventional frequency domain to the attenuated frequency domain.The inversion results could generally recover the missing amplitude and enhance the identification of weak signals.Besides,there were still residual interferences in the inversion results of the above processing method because of noise.Therefore,the inversion objective function was further optimized by introducing a Gaussian function based on the above method,which could effectively suppress the residual noised interferences.Then,high-resolution seismic data could be generated by convoluting the inversion results with broadband wavelet,thus,a thin reservoir identification method was formed based on compressed sensing with absorption and attenuation.The processing results of the theoretical model and the actual seismic data containing thin sand reservoirs show that compared with the conventional method,this method effectively enhances the weak signal amplitude and transverse continuity,and which can improve the resolution of seismic data and the identification ability of thin reservoirs while ensuring the signal-to-noise ratio.