一种高效的基于交叉梯度结构约束的三维地震走时与直流电阻率联合反演策略

摘要

基于交叉梯度结构约束的不同数据的联合反演可以提高地球物理成像的可靠度,但是由于不同观测数据对地下介质的灵敏度不同以及不同反演系统灵敏度矩阵元素的数值可能存在较大的差异,导致构建和求解联合反演系统存在很大的挑战.针对以上问题,本文提出一种新的基于单独反演模型更新量与交叉梯度结构约束相结合的联合反演策略.该策略利用单独反演系统分别确定出各个模型的更新量,然后利用它们约束交叉梯度系统的反演,得出新的模型更新量.通过这样的方式,有效实现了数据拟合与结构约束的平衡,实现了基于交叉梯度联合反演的目标.新的联合反演策略不需要对原来单独反演程序修改即可实现联合反演,减小了联合反演实现的难度,极大地提高了联合反演的易实现性,而且避免了联合反演矩阵存储及结构过于复杂难以求解的问题.基于新的策略,本文首次实现了基于交叉梯度结构约束的三维地震走时与直流电阻率联合反演.合成模型测试表明,与单一成像相比,联合成像减少了地震走时反演中出现的干扰异常并提高了电阻率反演的分辨率.%Joint inversion of different geophysical data based on cross-gradient structure constraint can improve the reliability of the imaging result.However,because different geophysical observations have varied sensitivities to the Earth medium properties,which are associated with different inversion systems,it is a big challenge to construct and resolve the joint inversion system.To solve the above problems,we propose a new joint inversion strategy by combining the model updates from separate inversions and cross-gradient structure constraint.This strategy first determines the model updates from separate inversions and then utilizes them to constrain the cross-gradient inversion system to get the new model updates.In this way,we can effectively implement the cross-gradient based joint inversion and realize the balance between fitting separate data types and increasing structure similarity between the two models.The new inversion strategy does not need to modify the individual inversion code to realize the joint inversion,making it much easier to conduct the cross-gradient based joint inversion.In addition,it also overcomes the difficulties in storing and inverting large joint inversion matrix.Based on the new strategy,for the first time we implement 3-D joint inversion of seismic travel-time and DC electrical resistivity data based on cross-gradient structure constraint.The test on a synthetic model shows that compared with separate inversions,the joint inversion method can effectively reduce the artifacts shown in seismic travel time inversion results and significantly improve the resolution of the resistivity inversion.