Foundation models for exploration geophysics
Qi Liu, Zenggui Chen, Jianwei Ma · Geophysics · 2025
ABSTRACT Recently, large models, or foundation models, have exhibited remarkable performance, profoundly impacting research paradigms in diverse domains. Foundation models, trained on extensive and diverse data sets, provide exceptional generalization abilities, allowing for their straightforward application across various use cases and domains. Exploration geophysics entails processing vast, multimodal, and multitask data sets that have traditionally relied on expert experience and physical principles. These unique characteristics present substantial challenges and compelling opportunities for advancing geophysical foundation models (GeoFMs). However, the development of GeoFMs in exploration geophysics is still at an early stage. This paper provides an overview of the current state and future prospects of GeoFMs in exploration geophysics. We begin by reviewing the development and emergent capabilities of foundation models, emphasizing their growing relevance in this domain. Furthermore, we discuss the hierarchy of GeoFMs for exploration geophysics and the critical techniques used, providing a research workflow that serves as a reference for their development. We then examine how GeoFMs may support various exploration tasks, with first-arrival picking based on the segment anything model as an example. Finally, we summarize the challenges faced in developing GeoFMs, along with future trends and their potential impact on the field.