3D Borehole to Surface Electromagnetic Inversion Based on Upscaling network
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
Borehole-to-surface electromagnetic method, due to its transmitter source being located in the borehole, can more directly excite electromagnetic responses from deep formations compared to surface electromagnetic methods, effectively improving the detection resolution of deep targets. However, for steel-cased wells, the casing possesses extremely high electrical conductivity, and the casing effect cannot be ignored in the data processing of borehole-to-surface electromagnetic surveys. Direct discrete modeling of slender hollow steel casings is both difficult and unfavorable for subsequent inversion calculations. Therefore, this paper proposes a three-dimensional borehole-to-surface electromagnetic inversion method based on an upscaling network. First, through unsupervised deep learning, the method establishes a nonlinear mapping between the electromagnetic response of fine-grid models containing casings and equivalent coarse-grid models, training an upscaling network from fine-scale grids to coarse-scale grid electrical structures. Second, the upscaled coarse-grid information is used as the initial model to perform three-dimensional borehole-to-surface electromagnetic inversion. Finally, the accuracy and feasibility of the proposed method are verified through three-dimensional borehole-to-surface electromagnetic inversion examples, and the inversion capability for deep targets is analyzed through synthetic data models, examining sensitivity regions, single-transmitter wells, and dual-transmitter wells.
Fields of Study & Subjects
Geophysical and Geoelectrical Methods
Geophysical Methods and Applications
Seismic Imaging and Inversion Techniques
Publication details
- Journal / source
- Academic Journal of Applied Sciences
- Volume
- 1
- Issue
- 1
- Pages
- 33–38
- Publication date
- 25 March 2026
- Language
- EN