Revolutionising ocean climate monitoring through space-based magnetometry and Artificial Intelligence

The tidal flow of ionized particles in seawater across Earth’s geomagnetic field induces electromagnetic (EM) signals within the ocean and solid Earth. These periodic signals are sensitive to oceanic properties such as salinity, temperature, and ultimately heat content, offering unique opportunities for climate monitoring. Despite 25 years of satellite-based tidal EM observations, their potential to estimate oceanic heat content (OHC) variations from space remains largely unexplored. PHOENIX proposes a pioneering framework to recover OHC from tidal EM fields by employing Physics-Informed Neural Networks (PINNs). This approach mitigates the inherent non-uniqueness of Earth sciences problems and the sparsity and uncertainties of available datasets by embedding physical constraints directly into the neural network architecture. A new 3-D EM solver will enforce these constraints and boundary conditions, enhancing the robustness and plausibility of the estimates. By combining advanced modelling with machine learning, PHOENIX will demonstrate how satellite magnetic data can be transformed into reliable OHC indicators. Systematic studies with realistic synthetic datasets will establish a framework for recovering physical properties from EM satellite data. A social media–focused scientific outreach strategy is envisaged, bringing fundamental climate concepts and scientific advancements to the general public. The techniques developed in this project will be readily applicable to the next generation of satellite-borne missions. This innovative approach will open a new avenue for global climate monitoring, strengthen interdisciplinary links between geophysics and climate science, and contribute to the EU’s efforts towards environmentally sustainable societal progress. The fellowship will equip the researcher with cutting-edge interdisciplinary expertise, boosting career prospects in both academia and applied climate innovation.