Shiji Cheng

Shiji Cheng

Auto-generated profile, based on publicly available information

Shiji Cheng

China University of Geosciences 1 School of Geophysics and Geomatics Wuhan, China; , Wuhan, China; and , Wuhan, China

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
9 January 2025 in Geophysics
Verified
Journal Article
Closed access
XWXinyu Wang
Xinyu Wang
HCHongzhu Cai
Hongzhu Cai
XHXiangyun Hu
Xiangyun Hu
LLLichao Liu
Lichao Liu
SLSHUANG LIU
SHUANG LIU
SCShiji Cheng
Shiji Cheng
Recent advancements highlight the benefits of unstructured tetrahedral meshes in addressing magnetotelluric (MT) modeling and inversion challenges. However, geoelectromagnetic inverse problems continue to face significant obstacles, including differing resolution requirements between forward modeling and inversion meshes as well as the blurring of conductivity boundaries in inversion results. To address these issues, we develop an innovative mesh decoupling strategy for forward modeling and inversion meshes. The subdomain decomposition method is used to decompose the inversion model into a series of curved-surface polyhedral subdomains, which act as inversion cells while preserving detailed topographic features. These subdomains are further subdivided into dense tetrahedral elements to ensure forward-modeling accuracy. In addition, to effectively obtain a sharper and more focused subsurface image, we integrate a zero-order minimum entropy (ZeroME) constraint in the objective function of 3D MT inversion. This constraint facilitates the acquisition of resistivity images with clear boundaries without necessitating the specification of appropriate focusing parameters, thereby providing greater flexibility. Synthetic and field data tests demonstrate that our method efficiently and reliably recovers subsurface conductivity distributions. The subdomain decomposition significantly reduces memory requirements and improves inversion stability, while the ZeroME constraint delivers more distinct model boundaries compared with traditional regularization methods.