Time Series Event Detection

Using Machine Learning to automatically detect ICMEs and other Structures

During my work at the Know Center, I contributed to various use cases for Time Series Event Detection within the Europlanet 2024 RI project:

  • Automatic Detection of Interplanetary Coronal Mass Ejections in Solar Wind In Situ Data
  • Automatic Detection of Magnetospheric Boundaries

In addition to a first-author publication (Rüdisser et al., 2022) and several conference contributions, this work also served as the basis of my master’s thesis.

The main approach is based on supervised machine learning and has been successfully applied to different datasets, such as Wind, STEREO-A or STEREO-B.

Event Detection

ARCANE

Based on previous work, I am currently working on the improvement of the ARCANE (Automatic Realtime deteCtion ANd forEcast) framework and integrating it into the real-time prediction pipeline at the Austrian Space Weather Office. ARCANE will serve as a real time monitor and contribute to automated operational forecasting. In collaboration with Gautier Nguyen at ONERA - The French Aerospace Lab, we are advancing the use of this methodology for early detection of ICMEs and prediction of key parameters.

Scores
Results per Event
ARCANE

This work is part of my PhD project, titled Combining AI and Physical Models to Advance Forecasting of Solar Coronal Mass Ejections, under the ERC project HELIO4CAST, which attempts to solve the Bz problem in heliospheric space weather forecasting.

References

2022

  1. Space Weather
    ruedisser2022.jpg
    Automatic Detection of Interplanetary Coronal Mass Ejections in Solar Wind In Situ Data
    Hannah T. Rüdisser, A. Windisch, Ute V. Amerstorfer, Christian Möstl , and 3 more authors
    Space Weather, Oct 2022
    DOI Bib