Generative AI for seismic data enhancement: a novel workflow using stable diffusion models

This research focuses on the application of Generative artificial intelligence (AI), specifically utilizing the state-of-the-art stable diffusion model, to generate and enhance 2D images of seismic amplitude maps. Seismic imaging is a critical tool in geosciences, particularly for assessing subsurface structures and evaluating the potential for carbon capture and storage (CCS). However, acquiring high-resolution seismic data can be both costly and logistically challenging, especially in areas where data is sparse or non-existent. To address this challenge, we propose a novel workflow leveraging generative AI to improve and create synthetic images of seismic amplitude maps based on existing analog data. This approach not only allows for the generation of new seismic images in data-limited regions but also enhances the resolution of existing images by computing high-quality representations from a compact latent space. This cost-effective and efficient solution allows us to improve the accuracy of subsurface characterizations and support more informed decisions in CCS initiatives.