Deep Learning Techniques for Inverting Observations from the EUV Snapshot Imaging Spectrograph

Imaging spectroscopy of the solar atmosphere with high spatial, spectral, and temporal resolution over a wide field of view is a longstanding goal of heliophysics because it allows for the measurement of important plasma parameters such as velocity and density with high fidelity. The EUV Snapshot Imaging Spectrograph (ESIS) is a sounding rocket instrument designed to capture EUV spectral line profiles over a large, 2D field of view with much higher temporal resolution than current rastering slit spectrographs. ESIS achieves this using a computed tomography imaging spectrograph design, with four channels. Each channel is an independent slitless spectrograph, illuminated by a common primary mirror, but oriented with a unique dispersion direction. Each ESIS exposure, comprising of four channels, can be inverted to recover spectral line profiles for every point in the field of view using limited-angle computed tomography techniques. In this work we present progress on the development of a deep learning algorithm that learns to solve the ESIS limited-angle tomography problem using data from the Interface Region Imaging Spectrograph as a training dataset. We will apply this algorithm to observations from the 2019 ESIS flight and compare the results to those obtained using previous methods.