Wave Velocity

Determining accurate seismic wave velocities is crucial for subsurface imaging and understanding Earth's structure, impacting resource exploration and hazard assessment. Current research heavily utilizes machine learning, employing generative models like diffusion models and transformers, as well as convolutional neural networks and encoder-decoder architectures, to improve velocity inversion from seismic data. These methods often incorporate prior information (e.g., well logs, geological structures) to enhance accuracy and address the ill-posed nature of the inverse problem. Improved velocity estimations lead to more precise subsurface models, benefiting various fields from oil and gas exploration to earthquake monitoring.