Physics-informed deep learning approach for modeling crustal deformation

• DOI: 10.1038/s41467-022-34922-1
• Publication Date: 2022-11-19T05:03:20Z
• Authors: Tomohisa Okazaki; Takeo Ito; Kazuro Hirahara; Naonori Ueda
• Keywords: Discontinuity (linguistics)

Abstract The movement and deformation of the Earth’s crust upper mantle provide critical insights into evolution earthquake processes future potentials. Crustal can be modeled by dislocation models that represent faults in as defects a continuum medium. In this study, we propose physics-informed deep learning approach to model crustal due earthquakes. Neural networks continuous displacement fields arbitrary geometrical structures mechanical properties rocks incorporating governing equations boundary conditions loss function. polar coordinate system is introduced accurately discontinuity on fault condition. We illustrate validity usefulness through example problems with strike-slip faults. This has potential advantage over conventional approaches it could straightforwardly extended high dimensional, anelastic, nonlinear, inverse problems.