Joint inversion of full-tensor gravity gradiometry data based on source growing
Hou Zhen-Long, Zhao Xin-Yang, Zhang Dai-Lei*, Zhao Fu-Quan, Wang Jia-Hui
1. School of Resources and Civil Engineering, Northeastern University, Shenyang 110819, China
2. Chinese Academy of Geological Sciences, Beijing 100094, China
3. China Solibase Engineering Co., LTD., Beijing 101300, China
Abstract：
Three-dimensional inversion based on source growing uses systematic searches. Compared with the regularization inversion, this method has lower computational requirements and faster processing speed. The criteria determining the source growth is crucial for the quality of the results. This study proposes an inversion based on source growing with full-tensor gravity gradiometry data to improve vertical inversion effectiveness. First, a depth weighting function is introduced for the criteria to optimize the determination of source growing at diff erent depths. Second, the weights of different data are adjusted based on the inversion results of singlecomponent gradient data, and a joint inversion method is established. Finally, matrix compression reduces memory occupation and improves computational efficiency. By the tests of synthetic data and real data from Vinton Dome, it is demonstrated that the proposed method can effectively guide source growing, providing stronger ability for distinguishing deep targets and being suitable for the inversion of complex-shaped targets. Furthermore, the method has high computational efficiency and anti-noise ability.
作者简介: Hou Zhen-Long, Post-doctor, got Bachelor’s degree of Geophysics from Jilin University in 2011; Ph.D. of Computer System and Architecture graduated from Jilin University in 2016; works as a postdoctor in Northeastern University from 2016 to 2019. The main research interests are gravity & magnetic exploration data processing & interpretation and the parallel computing methods. Address: No. 3-11, Wenhua Road, Heping District, Shenyang, P.R.China; Post Code: 110819
E-mail: houzlatjlu@163.com; houzhenlong@mail.neu.edu.cn
2024, Vol. 21 
|
