EGU25-12654, updated on 15 Mar 2025
https://doi.org/10.5194/egusphere-egu25-12654
EGU General Assembly 2025
© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.
Hybrid AI Approaches for Solar Feature Recognition Using Ground-Based Instrument Data
Oleg Stepanyuk1, Werner Pötzi2, Kamen Kozarev1, Momchil Dechev1, and Rositsa Miteva1
Oleg Stepanyuk et al.
  • 1Bulgarian Academy of Sciences, Institure of Astronomy, National Astronomical Observatory, Bulgaria (stepanyuk.oleg@aol.com)
  • 2University of Graz, Observatorium Kanzelhöhe, Austria, (werner.poetzi@uni-graz.at)
The dynamic behavior of solar prominences and filaments is a preursor to coronal mass ejections (CMEs), which can disrupt Earth's magnetosphere and affect satellite communications. Systematic ground-based solar observations, conducted with high temporal resolution, are instrumental in monitoring these structures. Analysis of the morphological changes and destabilization processes of filaments and prominences captured in datasets can help to identify early warning signs of potential eruptions. This capability is vital for developing reliable space weather forecasting systems, thereby mitigating the adverse effects of solar disturbances on Earth's technological infrastructure. Previously we introduced Wavetrack, a wavelet-based feature recognition software, which allowed, to a certain extent, to automate feature recognition for multiple events. We have since developed a convolutional neural network (CNN) model set which uses Wavetrack outputs as ground truth. Our initial model performance was shown on a set of SDO AIA instrument data performing segmentation of EUV and shock waves. In this work, we extend this hybrid approach for algorithmic and data-driven segmentation of on-disk solar features (prominences and filaments) using data from ground based-instruments, primarily focusing on Kanzelhöhe Observatory data. We discuss our approach to engineering training sets on real and synthetic data and the development of a CNN architecture generated within a general hyperparameter search routine. We showcase its performance on a set of filament/prominence events.

How to cite: Stepanyuk, O., Pötzi, W., Kozarev, K., Dechev, M., and Miteva, R.: Hybrid AI Approaches for Solar Feature Recognition Using Ground-Based Instrument Data, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-12654, https://doi.org/10.5194/egusphere-egu25-12654, 2025.

Comments on the supplementary material

AC: Author Comment | CC: Community Comment | Report abuse

supplementary materials version 1 – uploaded on 02 May 2025, no comments