Details of the Abstract
| Title of paper | Linearized model uncertainty analysis of 3D moderate- to large-scale MT inversion based on space transformation and efficient stochastic estimation |
| List of authors | H. Song, Y. Usui, T. Koyama, M. Uyeshima, P. Yu, K. Baba, B. Yang |
| Affiliation(s) |
State Key Laboratory of Marine Geology, Tongji University, Earthquake Research Institute, the University of Tokyo, Earthquake Research Institute, the University of Tokyo, Earthquake Research Institute, the University of Tokyo, State Key Laboratory of Marine Geology, Tongji University, Earthquake Research Institute, the University of Tokyo, Key Laboratory of Ocean and Marginal Sea Geology, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China |
| Summary | Geophysical inverse problems are unstable and non-unique. A meaningful solution should be composed of the preferred inversion model and its uncertainty estimates. However, almost all the current three-dimensional (3D) magnetotelluric (MT) inversion studies only offer the preferred model and ignore corresponding uncertainty estimates, which makes separating inversion artifacts from robust geological features impossible. One of the main bottlenecks is the huge computational cost, especially for large-scale 3D MT problems. Here, within the classical linearized model analysis framework, we try to demonstrate a relatively low-cost approach for accurately estimating the diagonal elements of the model covariance matrix by transforming the calculation domain from model space to data space with an efficient stochastic matrix diagonal estimator. The ability to estimate the diagonal of the covariance matrix thus facilitates the introduction of additional tools for model analysis, even for very large inverse problems, with storage and computational costs comparable to those required for obtaining inverse solutions. |
| Session Keyword | 2.0 EM theory, modelling and Inversion |
| File upload |
2.0_linearized_model_uncertai_song_05.pdf
|