多尺度快速匹配追踪多域联合地震反演方法

Seismic inversion in joint domain based on multi-scale fast matching pursuit

多尺度快速匹配追踪多域联合地震反演是一种通过地震数据多尺度分解的迭代反演方法.与此同时,在快速匹配追踪算法中引入低频模型约束,有效提高了收敛精度,使反演结果具有丰富的高低频信息.首先通过对大尺度地震资料进行反演得到低频背景.在此基础上,采用中尺度与小尺度地震数据进行逐级迭代用以获得高频数据,因而有效缓解了常规反演方法对于初始模型精度的依赖.最后利用理论模型与实际地震数据进行测试,通过与常规时间域反演方法的反演结果进行对比可以看出,本文方法在地层连续变化处依然可以对变化地层进行精确刻画,且在纵向分辨率提升的同时保持了较好的横向连续性.

Seismic inversion in joint domain based on multi-scale fast matching pursuit is an iterative cascade inversion approach via multi-scale decomposition of seismic data.At the same time,the low-frequency model constraint is introduced into the fast matching pursuit algorithm,which can improve the convergence accuracy effectively and make the inversion results with rich high-low-frequency information.At first,the low frequency background of subsurface media is obtained by large-scale seismic inversion.On this basis,the high frequency information of subsurface media is enhanced by intermediate and small-scale seismic inversion.Thus,the dependence of initial model accuracy can be relieved effectively by the proposed iterative inversion strategy.Finally,the 1-D model and the field data are processed by proposed method.The proposed method has much higher vertical resolution and better lateral continuity than conventional time-domain seismic inversion method.And the stratigraphic changes can still be well depicted at successive strata changes.