Speaker
Description
Machine learning (ML) has shown promise in space weather applications, yet its
predictive power is often limited by the scarcity of rare event data and the lack of
physical constraints. In this study, we explore a physics-informed neural network
(PINN) approach that integrates the VERB-CS model with a neural network model
to estimate cold plasma electron density in the plasmasphere. The network is trained
using in-situ observations from the Van Allen Probes (RBSP) and ARASE missions,
along with geomagnetic indices. Our goal is to assess whether embedding key physical processes such as particle transport, refilling, and loss mechanisms into the ML framework enhances forecasting performance, particularly during geomagnetically disturbed conditions.
We evaluate model skill across a range of geomagnetic activity, with a focus on the formation and evolution of plasmaspheric plumes at elevated Kp indices. Preliminary results suggest that the PINN approach captures both large-scale structure and
fine-scale features more accurately than purely data-driven models. Incorporating
physical knowledge into the machine learning model demonstrates increased generalizability across different geophysical conditions, including rare or extreme events. This work highlights the potential of combining physics-based models with data-informed learning to advance predictive capability in the near-Earth space environment.
| Do you plan to attend in-person or online? | In-person |
|---|
