基于光滑L0范数紧支撑聚焦的重磁联合反演的研究与应用

发布时间：2018-01-01 12:00

本文关键词：基于光滑L0范数紧支撑聚焦的重磁联合反演的研究与应用　出处：《吉林大学》2017年硕士论文　论文类型：学位论文

【摘要】：地球物理反演效果受诸多因素影响。首先,不同地球物理方法探测能力不同,进而导致不同方法反演结果垂向或横向分辨率存在差异。其次,由于实测地球物理数据受到人文干扰、自然条件、采集仪器精度等因素的影响,反演结果会出现不稳定的情况。此外,理论研究中便存在反演结果多解性的研究,即不同地下模型可能会表现出相同的数据响应特征。众多学者尝试对不同方法或者同一方法不同类型数据进行联合反演来提升反演效果,进而结合实际数据及地质、钻井资料,推断或者确定有关地质问题。然而,由于单一物性反演方法会严重影响联合反演效果,对单一及联合反演方法进行研究与改进,可以更好的对理论模拟及实际数据进行分析与讨论。本文将主要对重磁单一物性反演及联合反演方法进行研究与讨论。首先,给出反演过程所需的重磁核函数表达公式,通过设定三维剩余密度及剩余磁化率的模型空间分布,对模型重磁异常正演响应进行计算。其次,利用最小二乘反演方法进行模型物性反演计算。结果表明,该方法不适用。进而引入了正则化方法及对深度加权的讨论。通过构建基于光滑L0范数紧支撑聚焦正则化反演方法的目标函数,进行二维及三维模型正反演计算。引入交叉梯度函数,确立了重磁联合反演的目标函数及计算流程,对二维及三维模型进行正反演计算分析。最后,将基于光滑L0范数紧支撑聚焦方法应用于双湖地区剖面卫星重力数据反演,将基于该聚焦方法的联合反演方法应用于四川盆地及共和盆地的重磁数据反演计算,对结果进行了讨论分析。通过上述工作,可以得到如下成果及结论。首先,由于重磁核函数随深度衰减以及反演计算过程中的不适定问题,直接利用最小二乘方法的反演结果会出现明显的趋肤效应及多解结果,引入深度权函数及正则化反演方法可以较好的减少这些因素的影响。紧支撑聚焦作为正则化反演的一种方法,可以较好的突出异常体位置。而引入光滑L0范数的紧支撑聚焦方法,反演过程的计算效率及结果的聚焦效果均有提升。不同类型数据的交叉梯度函数值可以反映异常体不同属性值的变化情况,进而可以利用地质体的耦合关系,实现不同物性参数间的联合反演,且重磁联合反演同单一物性反演相比体现出了计算效率高,误差小等优点。利用光滑L0范数紧支撑聚焦方法对双湖地区剖面重力异常反演结果一定程度上体现出了缝合带两侧不同剩余密度岩体的分布情况,进而参考大地电磁数据反演结果及有关地质资料,推断出不同密度体的年代归属。应用基于光滑L0范数紧支撑聚焦联合反演方法对四川盆地剖面重磁异常数据反演结果展现出了剩余密度及剩余磁化率的分布情况,结合有关资料,给出了剖面基底不同岩性分布的推断。共和盆地重磁异常三维联合反演结果表现区域出了近似块状的深部密度异常体及近似铲状分布的磁化率异常体。本文通过模型模拟说明了基于光滑L0范数紧支撑聚焦的重磁联合反演方法的优势。通过对实际数据的探讨,进一步说明了该方法的有效性及实用性,为未来多重数据联合反演及更多实测数据应用提供参考。
[Abstract]:Geophysical inversion effect is influenced by many factors. First, different geophysical detection ability is different, which leads to the inversion results of different methods of vertical or horizontal resolution are different. Second, because the measured geophysical data by human interference, natural conditions, factors affecting the acquisition precision of instrument, the inversion results will become unstable. In addition, we research on the existence of multiple solutions of the inversion results in a theoretical study, namely different subsurface model may exhibit the same data. Many scholars try to response characteristics of different methods or the same method of different types of data inversion to improve the inversion effect, then combined with the actual geological data and drilling data, predict or determine the relevant geological problems. However, due to the single inversion method will seriously affect the inversion effect, the single and joint inversion method. The research and improvement, can better carry on the analysis and discussion of the theoretical simulation and actual data. This article will focus on the heavy magnetic single physical property inversion and joint inversion method is researched and discussed. Firstly, the heavy magnetic core function to be given in the process of inversion formula, by setting the 3D model space remaining density and residual magnetization rate the distribution model of gravity and magnetic anomaly forward response is calculated. Secondly, the model of inversion calculation using least squares inversion method. The results show that this method is not applicable. Then introduces the regularization method and discuss the depth weighted objective function. By constructing compactly supported smooth L0 norm regularization inversion method of focusing based on the two-dimensional and three-dimensional model inversion calculation. Introduce cross gradient function, established the objective function and the calculation process of joint inversion of gravity and magnetic, are against the two-dimensional and three-dimensional models Calculation analysis. Finally, the smooth L0 norm of compactly supported focusing method is applied to the profile of satellite gravity data inversion in Shuanghu area based on the inversion of gravity and magnetic data based on the application of the method of joint inversion of the focusing method in the Sichuan basin and the Gonghe basin calculation, the results were discussed and analyzed. Through the above work, we can get the following results and the conclusion. First, due to the heavy magnetic core function with depth attenuation and inversion calculation in the process of ill posed problem, inverse least squares method directly using the obvious skin effect and multi solution results, into the depth of weight function and regularization method can effectively reduce the influence of these factors a. Method of compactly supported focus as regularization, can better highlight the abnormal body position. Compactly supported focusing method introduced smooth L0 norm, efficiency calculation and inversion process The results were improved. The focusing effect of cross gradient function of different types of data values can reflect the changes of abnormal body of different attribute values, and then can use the coupling relationship between geological bodies, implementation of joint inversion of different material parameters and the joint inversion of gravity and magnetic properties compared with single inversion shows the advantages of high computation efficiency. Small error. Using the smoothed L0 norm of compactly supported focusing method for gravity anomaly inversion results in Shuanghu area to a certain extent reflects the suture zone on both sides of the distribution of different residual density of rock mass, and the reference ground electrical and magnetic data inversion results and related geological data, infer the different density in body ownership. Based on the application of smooth L0 norm the focus of compactly supported joint inversion method on the section in the Sichuan basin, gravity and magnetic anomaly data inversion shows the residual density and distribution of residual magnetization rate, According to relevant data, given different inferred lithological distribution basement. Gonghe basin, gravity and magnetic anomaly inversion results show that three-dimensional regional approximate block deep density anomaly and approximate shovel shaped distribution rate of magnetization anomaly. Through the model simulation show that the joint inversion method of gravity and magnetic focusing of the compactly supported smooth L0 norm based on the advantages. Through the discussion of the actual data, and further illustrates the effectiveness and practicability of the method, and provide reference for future joint inversion of multiple data and more data applications.

【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P631

【相似文献】