基于插值算法的大规模重磁数据快速协同反演策略

Joint inversion of multiple datasets and fast inversion of large-scale datasets are the research hotspots in the field of gravity and magnetic explorations. This project focuses on the two key issues of the joint inversion of large-scale gravity and magnetic datasets, including the joint methods of different parameters and the computation of large-scale datasets. We proposed a fast cooperative inversion method for gravity and magnetic data using the interpolation algorithm. For the joint method, a more general cooperative inversion scheme is proposed for data fusion, with which the difficult of multi-parameters coupling can be avoided. Moreover, this cooperative scheme is based on the inversion of single data. Therefore, it is convenient to incorporate appropriate constraints for the different dataset. For the computation of large-scale datasets, a fast modeling and inversion method is presented using the interpolation algorithm. Meanwhile, a technique for data compression is suggested based on the features of phase congruency, with which the data redundancy is further reduced to improve the computational efficiency. Combining the fast inversion method and the general cooperative inversion method introduced above, we implemented a fast cooperative inversion method of gravity and magnetic data. This project aims to promot the practicality of the joint inversion algorithms for gravity and magnetic data. With this project, the abilities of the gravity and magnetic methods for detecting the deep resources are further enhanced. The anticipated achievements are valuable both in theory and application.

多类型数据的联合反演和大规模数据的快速反演是当前重磁勘探领域的研究热点。本课题针对大规模重磁数据联合反演的两个核心问题——不同参数联合方式和大数据体快速计算，开展基于插值算法的大规模重磁数据快速协同反演策略研究。在联合方式上，研究提出一套适用性更广的协同反演框架进行数据融合，避开传统联合反演参数耦合的难题。在大数据体快速计算问题上，重点研究基于插值的快速正反演新方法。同时，提出基于相位一致性的数据压缩方法，进一步降低数据冗余以提高计算效率。课题预期研究目标是形成一套新的重磁联合反演快速计算方法体系，促进重磁联合反演算法的实用化，进一步提高重磁方法对深部资源的探测能力。预期研究成果具有重要的理论价值和应用价值。

本课题针对大规模重磁数据联合反演的两个核心问题——不同参数联合方式和大数据体快速计算开展研究。在联合方式上，本项目研究提出了一套适用性更广的协同反演框架进行数据融合，避开传统联合反演参数耦合的难题，在协同反演的框架中也尝试性引入了结构耦合来加强反演模型的结构相似性。在大数据体快速计算问题上，研究了基于插值和随机算法的快速正反演方法。课题形成了一套新的重磁联合反演快速计算方法体系，利用模型试验和实际数据试算验证了所提出算法的可行性和高效性。本项目的研究内容能够有效促进重磁联合反演算法的实用化，进一步提高重磁方法对深部资源的探测能力。