Tomographic Reconstructions of Coronal Mass Ejections with Physics-Informed Neural Radiance Fields

Coronagraphic observations enable the monitoring of coronal mass ejections (CMEs) through scattered light from free electrons. These observations allow for the estimation of the density, velocity, and propagation direction of the ejected plasma, which is critical for space weather forecasting. However, determining the 3D plasma distribution from 2D imaging data is challenging due to the optically thin medium and the complex image formation processes based on scattered light.

We present a method for 3D tomographic reconstructions of the heliosphere using multi-viewpoint coronagraphic observations. Our method leverages Neural Radiance Fields (NeRFs) to estimate the electron density in the heliosphere through a ray-tracing approach. The model is optimized by iteratively fitting the time-dependent observational data, accounting for the underlying Thomson scattering of image formation.  Typically, tomographic reconstructions based on a limited number of viewpoints are insufficient to constrain the 3D plasma distribution. To address this, we introduce additional physical constraints, including continuity, solar wind speed, and propagation direction, to enable a physics-informed tomographic reconstruction.

We utilize synthetic observations of CMEs based on GAMERA simulations to evaluate the model's performance with respect to viewpoint positions, physics-based constraints, and CME configurations. The results demonstrate that our method can reliably estimate the CME propagation direction and velocity using two viewpoints. Furthermore, we show that additional viewpoints can be seamlessly integrated, enhancing the reconstruction of the plasma distribution in the heliosphere and improving CME forecasting capabilities. This research underscores the value of physics-informed methods for 3D CME tomography, paving the way for advanced space weather monitoring.