Enhancing Observational Capabilities of EUV-observing Satellites to Estimate Spectral Irradiance

EUV irradiance strongly affects the density and temperature of the thermosphere, increasing or decreasing the atmospheric drag on low-Earth orbit (LEO) satellites. Our current best EUV irradiance forecasts are built to take advantage of EUV irradiance persistence. The challenge of this forecasting approach resides in the fact that the solar disk can change significantly within 1-2 days as new and old active regions rotate into and out of view. Here we present a novel approach to EUV irradiance forecasts that builds on the estimation of how the Sun looks from viewpoints along Earth’s orbit, that see the Sun that will later rotate into Earth’s perspective.

We present results that use SDO/AIA and STEREO-A & -B observations, as well as novel deep learning algorithms for the reconstruction of the 3D Sun called Neural Radiance Fields (NeRFs). Using this approach, we are able to generate viewpoints of the Sun that fall in between the viewpoints of STEREO and SDO. Data inferred from these novel viewpoints are used as input into a deep learning irradiance estimator trained on SDO/AIA and SDO/EVE observations to produce the EUV forecast.

Using this deep learning approach, we are able to forecast irradiance for observations of the Sun that do not currently have irradiance spectra associated with them. Our pipeline is an example of how novel deep learning techniques can be used to significantly enhance observational capabilities by the creation of virtual instruments.