Methodology and Acquisition for Multi-offset Ground Penetrating Radar: from Modelling to Advanced Multiscale full-waveform Inversion Applications.

• Applicant: Giacomo Roncoroni, Ph.D.
• Subject Area: Geophysics
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 575570494

The MAMMA MIA project aims to significantly enhance subsurface imaging in Arctic regions by developing advanced Multi-Offset Ground Penetrating Radar (MO-GPR) techniques. The main objective is to characterize fluid distributions in Arctic near-surface environments, vital for understanding permafrost stability and climate interactions. The project adapts established seismic exploration methodologies from the oil and gas industry. MAMMA MIA is divided into four integrated work packages (WPs): WP1 focuses on developing high-performance 2-D Full Waveform Inversion (FWI) and Reverse Time Migration (RTM) codes tailored for Arctic conditions, leveraging seismic inversion techniques successfully applied in industry. WP1 will deliver software validated through real-world Arctic GPR datasets, ensuring practical applicability. WP2 addresses the challenges of GPR data quality and inversion stability through advanced Deep Learning (DL)-based signal processing methods, together with classical algorithms. This includes developing DL models to enhance low-frequency data content, mitigate noise, and improve the accuracy of initial models. Techniques such as Long Short-Term Memory (LSTM) neural networks and convolutional models will be validated with field data to ensure robust performances. WP3 expands the developed 2-D methodologies into a scalable 3-D FWI framework, optimized for High-Performance Computing (HPC) infrastructures. The WP includes memory optimization strategies, such as wavefield reconstruction and compression techniques, to efficiently handle large datasets. WP4 develops and optimizes acquisition routines for MO-GPR. Initial dedicated field campaigns in Germany will provide essential datasets to validate each methodological advancement from WP1 to WP3, serving as preliminary validation before deployment in logistically complex Arctic environments. Acquisition setups will be rigorously tested and optimized for operational efficiency and high-resolution imaging. A final 3-D MO-GPR dataset will be acquired in Germany within this WP to validate and demonstrate the scalability and applicability of these advanced 3-D techniques. Successful project completion will result in open-source software tools, improved acquisition protocols, and scientifically validated inversion methods, significantly advancing Arctic subsurface exploration capabilities and supporting sustainable Arctic development.

• DFG Programme: Research Grants
• International Connection: Norway, USA
• Cooperation Partners: Professor Dr. Borge Arntsen; Professor Dr. Ettore Biondi; Professor Tor Arne Johansen; Professor Dr. Martin Landro