INTELLIGENT COMPUTING FOR SEISMIC INVERSION WITH AUTOMATIC DIFFERENTIATION
Seminars
Semester 2
Understanding Earth's subsurface structure is essential for resource exploration and hazard assessment. Traditional seismic inversion methods rely on adjoint-state methods for gradient computation. It faces some major limitations: complex manual derivations and poor flexibility across different wave physics. These barriers significantly impact research efficiency and limit our ability to understand deep Earth structures. We propose a novel framework using intelligent computing with automatic differentiation (AD) for seismic inversion. AD, a modern computational technique that automatically computes exact derivatives through chain rule decomposition, offers a promising solution. Our research first focuses on establishing mathematical equivalence between AD-computed and traditional gradient computations. The planned framework will support multiple wave types - Acoustic, Love, and Rayleigh waves. To optimize memory usage in large-scale inversions, we will implement an adaptive quadtree structure. Different field data require unique intelligent preprocessing approaches. Here we will first develop a semi-supervised signal extraction method for Distributed Acoustic Sensing data as an example. Through this framework, we aim to improve research efficiency and enable comprehensive multi-parameter joint inversions, ultimately advancing our understanding of Earth's structures through more effective seismic imaging.
Additional information: Mr. Xi WANG, u3009920@connect.hku.hk