Authors: Jaime A. Landeros (Department of Mechanical and Aerospace Engineering, University of California San Diego), Michael S. Kirk (NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA), C. Nick Arge (NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA), Jeremy A. Grajeda (Klipsch School of Electrical & Computer Engineering, New Mexico State University, Las Cruces, NM 88003, USA), Laura E. Boucheron (Klipsch School of Electrical & Computer Engineering, New Mexico State University, Las Cruces, NM 88003, USA), Jie Zhang (George Mason University, Fairfax, VA 22030, USA)

The segmentation of Coronal Holes (CHs) from solar imagery plays a pivotal role in constraining both numerical models of the corona and theories on solar wind formation at CH boundaries. We present a method for CH segmentation with Uncertainty Quantification (UQ) from coronal Extreme Ultraviolet (EUV) 193 Å images, photospheric magnetograms inferred from visible 6301.5 Å, and chromospheric near-infrared He I 10830 Å images. It builds on an EUV-based and a He I-based method to address the challenges inherent in 2D, monochromatic measurements of the 3D, multi-thermal solar atmosphere. This new approach demonstrates enhanced binary classification performance, more reliable uncertainty estimates, improved accuracy in identifying polar CH boundaries, and earlier as well as longer-lasting detections as the Sun rotates, compared to the two constituent methods. The maturation of such data-derived constraints improves their utility towards the calibration of global coronal models and the evaluation of solar wind formation theories.