随着计算机技术的飞速发展,大地电磁测深数据的各种反演方法得到空前发展.对于反演过程中的网格剖分问题,前人研究有限.本文就此进行研究,确定影响网格剖分方式的因素,找到一种合适的网格剖分方法.为此,设计一个复杂模型,对正演模拟得到的数据加入随机噪声作为反演的输入数据,选择不同的网格剖分方式,使用DASOCC与NLCG反演方法,以相同参数分别进行反演,对所得结果进行对比分析.结果显示,在设定的模型及参数条件下,针对DASOCC反演方法,纵向及横向网格均不加密,反演效果比较理想,即横向采用一个测点对应一个网格的剖分方式,纵向采用表层加密,100 m以下后一网格为前一网格厚度1.1倍的剖分方式;针对NLCG反演方法,纵向与DASOCC反演方法相同,采用不加密网格的剖分方式得到的结果较好,而横向则建议采用一个测点对应一个网格,并在2个测点之间插入一个网格的剖分方式.%With the development of computer technology, magnetotelluric data inversion method has been an unprecedented development. A very important issue in the process of inversion is how to generation mesh, and which standards can be follow. Previous studies on this issue are very limited. In this paper, the study on this issue can help us to determine the factors that affect the mesh generation, and ifnd a suitable mesh generation method. So we design a complex model, and add the random error to the data, which get from the forward, as the inversion input data. And then the data is inversed whit the different mesh generation method, the different inversion methods, and the same parameters. The results obtained are compared and analyzed. The results show that, under the conditions of the model and parameters which be used in this paper, the DASOCC inversion, when the vertical and horizontal mesh all get less, can get the satisfactory results. That is, using the mesh of one lattice for one measuring point in horizontal and gradually more sparse lattice in vertical, 100m in the surface and 1.1 times in turn to depth. For a better inversion result of the NLCG inversion, it is suggested the same mesh in vertical as DASOCC inversion, but inserting a lattice between two measuring points in horizontal.
展开▼
