Graph-based Neural Space Weather Forecasting
Daniel Holmberg, Ivan Zaitsev, Markku Alho, Ioanna Bouri, Fanni Franssila, Haewon Jeong, Minna Palmroth, Teemu Roos
TL;DR
The paper addresses the need for fast, uncertainty-aware space weather forecasting by learning a graph-based emulator trained on high-fidelity hybrid-Vlasov simulations (Vlasiator). It introduces deterministic Graph-FM and probabilistic Graph-EFM architectures that operate on simple, multiscale, and hierarchical mesh graphs to autoregressively predict near-Earth conditions driven by upstream solar wind. The probabilistic model uses a latent-variable approach to generate ensembles, enabling uncertainty quantification via CRPS calibration, while the deterministic variant provides rapid point forecasts. Results show substantial speedups (up to ~500x) over full Vlasiator runs and demonstrate the potential of graph-based emulation to deliver real-time, uncertainty-aware space weather forecasts, with future work focusing on 3D extensions, larger datasets, and physics-constrained improvements.
Abstract
Accurate space weather forecasting is crucial for protecting our increasingly digital infrastructure. Hybrid-Vlasov models, like Vlasiator, offer physical realism beyond that of current operational systems, but are too computationally expensive for real-time use. We introduce a graph-based neural emulator trained on Vlasiator data to autoregressively predict near-Earth space conditions driven by an upstream solar wind. We show how to achieve both fast deterministic forecasts and, by using a generative model, produce ensembles to capture forecast uncertainty. This work demonstrates that machine learning offers a way to add uncertainty quantification capability to existing space weather prediction systems, and make hybrid-Vlasov simulation tractable for operational use.