低频稀疏双约束宽频带地震阻抗反演

传统波阻抗反演方法求解的反射系数没有反映足够的低频信息,其连续性、分辨率存在着一定的局限性,严重影响了波阻抗反演的质量和分辨率。为此,以理论合成记录和子波模拟为支点,讨论了富低频信号的优点;引入低频稀疏双约束反射系数方法进行稀疏优化项的改进,形成了新的方法 ;并基于实际工区的宽频带资料和非宽频带资料,应用传统的基追踪反射系数反演进行反射系数、波阻抗反演对比,再与低频稀疏双约束反射系数反演求解的反射系数、波阻抗反演进行对比,以验证改进方法的效果。研究结果表明:①具有丰富低频信息的宽频数据受旁瓣的影响更少,其地震数据分辨率更高,更有利于反演精度和分辨率的提高;②新方法在BPDN基追踪降噪问题的基础上增加了L2范数低频模型约束残差项,可实现含低频信息反射系数的直接求解;③较之于传统方法,新方法所求解的反射系数、波阻抗具有更好的连续性和分辨率,并且能与井数据良好吻合。结论认为,新方法在宽频带数据和非宽频带数据的波阻抗反演上得到了更高的分辨率,提高了波阻抗反演的准确率,在薄储层分布预测中具有更高的应用价值。

Low-frequency sparse double-constrained broadband seismic impedance inversion

The reflection coefficient solved by using the traditional impedance inversion method cannot reflect low-frequency information sufficiently and its continuity and resolution are limited, so the quality and resolution of impedance inversion is impacted seriously. In this paper, the advantages of rich low-frequency signals were discussed with the theoretical synthetic seismogram and wavelet simulation as the fulcrum. Then, the low-frequency sparse double-constrained reflection coefficient method was introduced to modify the sparse optimization item, and thus a new method was formed. Finally, based on the broadband data and non-broadband data of the actual work area, the reflection coefficient and impedance inversion solved by the traditional basis-pursuit reflection coefficient inversion were compared and then the calculation results of low-frequency sparse double-constrained reflection coefficient inversion were compared to verify the effect of the modified method. And the following research results were obtained. First, the broadband data with rich low-frequency information is less affected by the side lobe, and its seismic data resolution is higher, which is more favorable for the improvement of inversion accuracy and resolution. Second, in the new method, L2 norm low-frequency model is added on the basis of BPDN basis-pursuit denoising problem to constrain the residual, so as to realize the direct solution of the reflection coefficient with low-frequency information. Third, the reflection coefficient and wave impedance solved by using the new method have better continuity and resolution than those solved by using the traditional method and they are in good agreement with the well data. In conclusion, the new method achieves higher resolution on the impedance inversion of broadband data and non-broadband data and the accuracy of impedance inversion is increased, so it has a higher application value in predicting the distribution of thin reservoirs.