中文摘要：

Occam反演是一种正则化的光滑模型反演方法,它在寻找最小拟合差的同时追求最光滑模型.因地下介质的电性通常是连续变化的,Occam反演考虑了地下介质横向和纵向的光滑情况,应用Occam反演避免了模型电性参数被随意间断.通过建立层状和存在两个异常体的模型,利用2.5维可控源音频大地电磁有限元法作正演响应计算,对响应数据基于Occam反演理论拟合反演.结果表明,Occam反演对可控源音频大地电磁法响应数据反演稳定收敛,理论模型与反演结果的视电阻率曲线形态基本一致,测点数据的拟合误差小于3％,反演结果能反映出模型电阻率的光滑效果.

英文摘要：

Occam inversion is a regularization inversion to generate smooth models. The smoothest model is sought to the criterion that minimizes the misfit to the data. As resistivity of underground medium changes continually, Occam inversion considers both lateral and vertical roughness of underground medium. It avoids resistivity of model be interrupted arbitrarily. The paper takes a layered model and with two abnormities models for example, uses finite element methods of 2. 5 dimensional controlled source audio-frequency magnetotellurics to calculate response, then finds the models that fitting the response data by Occam inversion method. The results show that Occam inversion to CSAMT data is stable and convergent, the app-resistivity lines with high coherence, the misfit of a set data less than 3%, and the smoothness is presented in inversion results.