Coronal jet Identification with machine learning

¹ Solar Physics and Space Plasma Research Centre (SP2RC), School of Mathematical and Physical Science, University of Sheffield, Hicks Bldg, Hounsfield Road, Sheffield S3 7RH, UK
² Department of Physics, University of Rome “Tor Vergata”, Via della Ricerca Scientifica 1, Rome I-00133, Italy
³ Instituto de Astrofísica e Ciências do Espaço, Department of Physics, University of Coimbra, Coimbra, Portugal
⁴ Department of Astronomy, Eötvös Loránd University, Pázmány P. sétány 1/A, Budapest H-1117, Hungary
⁵ Gyula Bay Zoltan Solar Observatory (GSO), Hungarian Solar Physics Foundation (HSPF), Petőfi tér 3., Gyula H-5700, Hungary
⁶ National Key Laboratory of Deep Space Exploration, School of Earth and Space Sciences, University of Science and Technology of China, Hefei 230026, People’s Republic of China
⁷ CAS Key Laboratory of Geospace Environment, Department of Geophysics and Planetary Sciences, University of Science and Technology of China, Hefei 230026, People’s Republic of China
⁸ School of Electrical and Electronic Engineering, University of Sheffield, Amy Johnson Building, Portabello Street, Sheffield S1 3JD, UK

^⋆ Corresponding author: s.chierichini@sheffield.ac.uk

Received: 19 September 2024
Accepted: 5 March 2025

Abstract

Coronal jets are narrow eruptions observable across various wavelengths, primarily driven by magnetic activity. These phenomena may play a pivotal role in solar activity, which significantly impacts the dynamics of the solar system, however they have not been studied in depth thus far. This work employs machine learning, specifically, via a random forest model, to enhance the assembly of the dataset of coronal jets. By combining data from two segmentation methods, semi-automated jet identification algorithm (SAJIA) and mathematical morphology (MM), we strove to develop a more comprehensive dataset. Our model was trained and validated initially on a robust dataset and subsequently applied to classify unlabelled data. To ensure a higher level of confidence for positive identifications, the classification threshold was increased to 0.95. This adjustment led to the identification of 3452 new jet candidates. The new candidates were then validated through visual inspection. The validation resulted in the identification of 3268 true jets and 184 false positives. Our findings highlight the effectiveness of integrating machine learning with traditional analysis techniques to enhance the accuracy and reliability of solar jet identification. These results contribute to a deeper understanding of coronal jets and their role in solar dynamics, demonstrating the potential of machine learning in advancing solar physics research.