基于一维反演结果引导的大地电磁三维反演研究

Research on magnetotelluric three-dimension inversion guided by one-dimension inversion results

合理构建初始模型是当前大地电磁三维反演实用化中迫切需要解决的难题.然而,传统初始模型构建方法存在费时和不客观的缺点,造成反演解释的困难.为此,我们提出基于一维反演结果引导的初始模型构建方案,通过对不同数据集的一维反演结果进行三线性插值,并提取全区最小Bostick深度的相似信息以构建三维反演的初始模型,进而更好地恢复地下结构.利用该方案对一系列理论数据进行反演试算,结果表明基于一维反演结果构建初始模型可有效改善三维反演的效果,特别是对异常界面的刻画更接近真实模型.鉴于Occam一维反演方法构建初始模型相比于其他方法具有反演速度快和结果可靠的优势,采用该方案对内蒙古塔木素地区大地电磁实测数据进行三维反演试算,其结果与该地区地震解释结果吻合良好,更符合野外地质情况,验证了本文提出方案的实用性.本文提出的大地电磁三维反演初始模型构建方案可为实测数据处理及解释提供借鉴和参考.

How to reasonably construct the initial model is an urgent problem to be solved in the practical application of magnetotelluric 3D inversion.However,the traditional initial model construction method is time-consuming and non-objective,which makes the inversion interpretation difficult.Therefore,we propose an initial model construction scheme based on 1D inversion results.Through trilinear interpolation of 1D inversion results from different datasets and extracting similar information of the minimum Bostick depth in the whole region to construct the initial model of 3D inversion,so as to better restore the underground structure.Using this scheme,several theoretical models are inversed,the results show that the initial model based on 1D inversion results can effectively improve the effect of 3D inversion,especially the description of abnormal interface is closer to the real model.In addition,compared with other methods,the initial model based on Occam method has the advantages of fast inversion speed and reliable results.On this basis,the Occam 1D inversion results of magnetotelluric data in Tamusu area are used to construct the initial model and complete the 3D inversion.The result is in good agreement with the seismic interpretation result in this area,which is more consistent with the field geological conditions,which verifies the practicability of the scheme proposed in this paper.The initial model construction scheme of magnetotelluric 3D inversion proposed in this paper can provide reference and reference for measured data processing and interpretation.