Wellbore-supervised learning for 3D seismic velocity model building
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
The near surface, particularly in arid regions, poses significant challenges for seismic exploration due to extreme velocity variations. Traditional refraction travel-time-based inversions struggle to reconstruct shallow velocity reversals making it crucial to develop alternative solutions. Uphole surveys, though valuable for estimating weathered layer and formation velocities, remain largely underutilized in modern velocity modeling practices. Our objective is to combine scattered uphole velocity calibrations with 3D surface seismic data to build accurate 3D velocity models. By leveraging machine learning's data-driven regression capabilities, we bypass complex physical modeling, enabling efficient and reliable reconstruction of 3D near-surface velocity models using multi-wave arrivals recorded in the seismic gather. The novel approach is demonstrated on the synthetic SEAM arid model and on field data acquired in complex near surface geology.
Fields of Study & Subjects
Seismic Imaging and Inversion Techniques
Seismic Waves and Analysis
Seismology and Earthquake Studies
Publication details
- Journal / source
- Second International Meeting for Applied Geoscience & Energy
- Volume
- 1
- Pages
- 1637–1641
- Publication date
- 10 February 2026
- Language
- EN