| Abstract: Understanding and predicting the evolution of space weather drivers (solar wind, CME, solar flare) remains a central challenge due to the multiscale and coupled nature of heliophysical processes. This talk presents a unified modeling framework developed during my PhD to address this challenge through physics-based and data-driven approaches.
First, I introduce the SWASTi framework, a 3D MHD-based solar wind model that provides self-consistent background conditions by coupling coronal magnetic field extrapolations with heliospheric evolution. Building on this, I present SWASTi-CME, a physics-based model that captures CME evolution under varying ambient solar wind conditions, highlighting the role of drag and internal magnetic structure. I then discuss CME-CME interactions and their impact on geo-effectiveness, demonstrating how interactions can enhance disturbances and modify heliospheric structures. Complementarily, a machine learning approach using gradient boosting is explored for solar flare prediction, emphasizing feature selection and forecasting skill.
Finally, I outline ongoing work on hybrid physics+AI models, including WSA+ enhancements and MHD suuroagte based on physics-informed neural operator, aimed at enabling fast, accurate, and uncertainty-aware space weather forecasting. |
