结合实际数据的三维可控源电磁法正反演研究

发布时间：2018-06-06 06:01

本文选题：可控源电磁法 + 三维正演　；参考：《吉林大学》2015年硕士论文

【摘要】：可控源音频大地电磁测深(CSAMT)自上世纪七十年代以来，已经不断发展为一种重要的电磁法勘探手段并得到广泛的应用。随着CSAMT法的理论发展，及电磁勘探仪器方面的研究方面的进步，对数据处理方法和解释均提出了更高的要求，迫切地需求三维数据的采集以及精确有效的数据处理方法，针对这一现状，本文对理论数据进行三维数值模拟并进行有限内存拟牛顿法反演，然后结合可控源电磁勘探实际数据进行研究。本文利用交错网格有限差分法对可控源音频大地电磁法的理论数据进行三维正演。对于层状背景模型中存在低阻异常体和高阻异常体的情况的电场分布特点进行了分析和总结。反演方面，采用的是有限内存BFGS法(L-BFGS)。该方法是拟牛顿法的一种，不需要存储大型海森矩阵或它的逆矩阵，只需存储前面m次的迭代信息，从而减小了反演时占用的计算机空间。利用有限内存BFGS法对文中正演部分的理论模型计算结果，通过给出不同的反演初始背景电阻率对不同频率的数据进行反演，并对结果进行分析。反演结果说明，该方法对于理论数据可以准确反映异常体的电阻率高低、位置、形态的信息，证明了有限内存拟牛顿法对进行三维可控源频率测深数据反演的可行性。最后，本文对实测的可控源数据进行了反演模拟。将实测数据编辑之后进行反演，为了计算方便，，选择单个频率进行反演。从实际数据的反演结果中可以看出，由于实际地质情况更加复杂，只能根据反演的电阻率分布情况判断地下结构，而反演的电阻率数值不够精确，因此需要在实际数据的反演方面进行更加全面和深入的研究。本文通过对理论模型和实际数据正反演计算分析得出以下结论：（1）利用有限差分交错网格法对理论模型进行正演，对计算结果进行对比分析可以总结出不同地质条件下电场的响应规律，通过对响应规律的总结可以为反演工作提供思路和参考；（2） L-BFGS法三维反演在理论模型中的应用说明，选择合适的反演参数，该方法可以比较准确的反映出异常体的位置形态及电阻率的高低，拟合差下降速度快，模型数据收敛明显；（3）对实测数据进行了L-BFGS法三维反演。实际数据反演结果中高阻低阻区分明显，可以根据电阻率的分布情况判断地下地质构造情况。
[Abstract]:Since the 1970s, the controllable source audio magnetotelluric sounding (CSAMT) has been developed as an important method of electromagnetic exploration and has been widely used. With the development of the theory of CSAMT method and the progress of the research on electromagnetic exploration instruments, the data processing method and interpretation are required to be higher, and the 3D data acquisition and accurate and effective data processing methods are urgently needed. In view of this situation, this paper carries on the three-dimensional numerical simulation to the theoretical data and carries on the finite memory quasi Newton method inversion, then unifies the controllable source electromagnetic exploration actual data to carry on the research. In this paper, the staggered grid finite difference method is used to forward the theoretical data of controllable source audio magnetotelluric method. The characteristics of electric field distribution in the case of low resistivity anomaly and high resistivity anomaly in layered background model are analyzed and summarized. In inversion, the finite memory BFGS method is used. This method is a quasi-Newton method, which does not need to store large Heyson matrix or its inverse matrix, but only needs to store the information of the previous m iterations, thus reducing the computer space used in inversion. The finite memory BFGS method is used to calculate the theoretical model of the forward part of the paper. The data of different frequencies are inversed by different inversion background resistivity, and the results are analyzed. The inversion results show that the method can accurately reflect the information of resistivity, position and shape for theoretical data, and proves the feasibility of the finite memory quasi-Newton method for inversion of 3-D controllable source frequency sounding data. Finally, the inversion simulation of the measured controllable source data is carried out. After editing the measured data, the inversion is carried out, and for the convenience of calculation, a single frequency is selected for inversion. From the inversion results of the actual data, it can be seen that because the actual geological conditions are more complicated, the underground structure can only be judged according to the resistivity distribution of the inversion, but the inversion resistivity value is not accurate enough. Therefore, more comprehensive and in-depth research on the inversion of actual data is needed. In this paper, the following conclusions are drawn through the analysis of the theoretical model and the actual data forward and inverse calculation. 1) by using the finite difference staggered grid method to forward the theoretical model, the response law of electric field under different geological conditions can be summed up by comparing and analyzing the calculated results. Through the summary of the response law, we can provide ideas and references for the inversion work. 2) the application of L-BFGS method in theoretical model shows that by selecting appropriate inversion parameters, this method can accurately reflect the position shape and resistivity of abnormal body, and the falling speed of fitting difference is fast, and the convergence of model data is obvious. 3) three dimensional inversion of measured data by L-BFGS method is carried out. The difference between high resistivity and low resistivity is obvious in the inversion result of actual data, and the underground geological structure can be judged according to the distribution of resistivity.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.325

【参考文献】