| Abstract |
The project pursued two objectives: developing a fully remote, desktop-based subsurface imaging method using natural atmospheric muons—eliminating on-site hardware and minimising costs—and creating an automated simulation workflow to optimise detector type, placement, energy range, and environmental conditions for underground settings. Together, these aims demonstrate that muon tomography can be entirely virtual, establishing the foundation for a precise new imaging technique.
The project set a control site at the SURF mine in South Dakota. Geo-environmental mapping, 3D modelling, and a Geant 4 pilot simulation were developed. Muon data and geo-environmental variables were added to the system. Multiple validations, based on visualisation of the data, were performed on the SURF mine. The project was further enhanced to a full-stack system design, including the required software development and code script enhancements. Machine learning on 5 million custom data sets was carried out, resulting in an advanced, intelligent AI system. The output from a MANTIS simulation sources from twenty-six different data reference feeds, such as inSAR, GRACE and Sentinel 2. Final system enhancements included a simple user interface, data modelling, and full validation across several non-test sites.
The project confirms muon-based subsurface imaging is viable and fully computer-simulated, establishing a new technique. Simulations showed that optimal detector placement requires surrounding targets with high-resolution, low-noise sensors. Validated real-site tests prove remote modelling of muon interactions can reveal hidden structures. This groundbreaking, non-invasive, cost-effective method has wide applications in mining, archaeology, construction, and national security.
|