Training of prediction of ICME speed and arrival time by machine learning

Student: Michal Zummer
Vedoucí: Gilbert Pi, Ph.D.
Konzultant: Doc. RNDr. František Němec, Ph.D.
Stav projektu: dokončený

Anotace:

Determining the exact arrival time of an interplanetary coronal mass ejection from the Sun to the Earth is one of the greatest challenges of actual space physics. In this project, we will use machine learning to predict an ICME² arrival time, maximum speed, and Dp³ in Wind observations by the SOHO/LASCO⁴ recorded CME¹ parameters.

LASCO recorded parameters of more than 700 halo CMEs (https://cdaw.gsfc.nasa.gov/CME_list/halo), and the corresponding ICMEs are observed by the Wind satellite at the L1 point (https://wind.nasa.gov/ICME_catalog). The student needs to review these two lists, find the related events, and then apply the deep learning method to regression. The Python code for regression can be found on the website⁵.

1. CME = Coronal Mass Ejection
2. ICME = Interplanetary Coronal Mass Ejection
3. Dp = dynamic pressure
4. SOHO/LASCO (https://lasco-www.nrl.navy.mil)
5. Python regression method (https://machinelearningmastery.com/regression-tutorial-keras-deep-learning-library-python/?fbclid=IwAR0JoahCw2opv2Iq4lfxmk467WZlLJwyiyx_6kYEY3BRbUb8AU519g5GTUs)