Research And Application Of Pseudo Two-Dimension Inversion Of Magnetotelluric With Lateral Constraint

At the present, although the two-dimension and three-dimensional magnetotelluric inversion can take the partition of underground structure accurately. While due to the process of the data of many dimensions and the algorithm complexity, the calculation of the inversion is very slow. On the other hand, the computation takes up lots of memory and has long time-consuming, so there is no advantage of getting the instruction of underground during actual data interpretation. MT one- dimensional inversion has fast speed and small demand of memory, it can give a preliminary explanation of field data materials very fast. As the advantage mentioned above, this paper have took the pseudo-two-dimensional inversion of multiple data points.In the calculation process, the horizontal and vertical restraints had been added to the model to avoid the inversion result which between adjacent measuring points mutations.Firstly, this paper built the K and H type model to take forward simulation.Inversing the data which is added to the noise to verify the L-BFGS algorithm.Secondly, this paper studied inversion effects of different factors,such as the thickness 、 the resistivity distribution 、 regularization factor 、 Jacobi matrix. Otherwise, this paper used staggered grid finite difference method to simulate the high resistivity body and low resistivity body which generate forward data. I inverted the data with the lateral constraint of model, giving the data fitting result and iteration condition. Last, this paper use the pseudo two-dimension inversion method with lateral constraint to invert the real data of Tan Lu region, and made an overall comment of the inversion result.From the study of this paper, we can get the following conclusions:1、We take MT one-dimensional inversion use L-BFGS method can obtained a good reflection of the structure of layer.2、Whether choose the initial models which is uniform thickness or layer thickness increase,the inversion result will be good.3、The regularization factor usually set to be 3. Too big or two small will cause the inversion effect distorition.4、Using the analytical method to calculate the sensitivity matrix can greatly reduce the times of forward, shortening the inversion time and improve efficiency. At the same time,the Jacobi elements can reflect more sensitive of deep structure. 5、The inversion result of 2D and 3D forward data can reflect the shape of anomalies. 6 、 The comparison of the one-dimensional inversion, pseudo Two-dimension inversion result and SCS2 D software inversion result show that the one-dimensional inversion showd a banded distribution and there are false anomalies in some degree. Although there is different result of local area, the overall reflection of the whole underground structure is similar. It illustrated that the MT pseudo two-dimension inversion with lateral constraint can take a preliminary judgment of deep structure and provide a reference for the precessing and interpretation of field data.