Project Summary

The project, conducted as part of my PhD, aimed at integrating data-driven methodologies with physical models to capture the dynamics of geomagnetic storms.

• It was developed a framework combining traditional deterministic models with Physics- Informed Neural Networks (PINNs) to model the temporal evolution of the geomagnetic SYM-H index
• It was used high temporal resolution (1-minute) data from OMNIWeb, including solar wind plasma and interplanetary magnetic field measurements, to study rapid magnetospheric variations.
• Interactions with ESA Φ-lab through literature studies, interactive discussions, and technical presentations, thereby broadening my knowledge and understanding of AI applications in Earth Observation.

Development Tools

• Used Python for all code development of practical implementations of PINNs.
• Extracted high-resolution SYM-H index data and relevant solar wind parameters from OMNIWeb (https://cdaweb.gsfc.nasa.gov/).

Development Outputs

• All developed codes, including implementations and examples, are maintained on a private GitHub repository.
• Presentations at Φ-lab.
• Processed datasets used in the analysis are archived with metadata descriptions, ensuring reproducibility.