Data-driven segmentation of solar filaments based on ground-based instrument data
Abstract
The dynamic behavior of solar prominences and filaments is a precursor 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 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 hybrid algorithmic-data driven method which performance was shown on a set of Solar Dynamic Observatory (SDO) Atmospheric Imaging Assembly (AIA) instrument data performing segmentation of extreme ultraviolet (EUV) and shock waves. In this work, we extend certain aspects of this approach to image segmentation of solar features -- filaments and prominences using H-alpha Kanzelhohe Observatory data. We describe a leakage-aware, sparse multi-year sampling strategy. We present image segmentation models pre-trained on synoptic and normalized image data, followed by a discussion of how model performance depends on training-set engineering, data preprocessing techniques, and the choice of loss metrics. We perform a limited study, both analytically and numerically, of the effect of external, fixed normalization versus native input for compact U-Net-based segmentation of solar H-alpha filaments, also contributing to ongoing research on normalization in deep networks in general and compact models in particular.
Versions
Related
- Morphology of Solar Filaments During the Solar Maximum of Cycle 24: A Study of 35 Filaments Focusing on Chirality, Barb Orientation, Skewness, and Eruptive Behavior
- Multi-Instrument Analysis of NOAA AR 12781: Coupling Surface Evolution of the AR with Its In-Situ Solar Wind Signatures
- Spatially variant neural blind deconvolution to reconstruct ground-based solar observations
- Effect of Gravity-driven Longitudinal Flows in Filaments on Angular Momentum Transport to Embedded Cores
- SoFTg: A hybrid segmentation framework for solar granulation revealing turbulent convection statistics at DKIST scales
BibTeX
@misc{stepanyuk2026datadriven,
title = {{Data-driven segmentation of solar filaments based on ground-based instrument data}},
author = {Oleg Stepanyuk and Werner Pötzi and Momchil Dechev and Rositsa Miteva and Kamen Kozarev},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.34502},
eprint = {2610.34502},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.34502},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Data-driven segmentation of solar filaments based on ground-based instrument data
AU - Oleg Stepanyuk
AU - Werner Pötzi
AU - Momchil Dechev
AU - Rositsa Miteva
AU - Kamen Kozarev
PY - 2026
DA - 2026/10/25
PB - arXiv
DO - 10.48550/arxiv.2610.34502
UR - https://arxiv.org/abs/2610.34502
AB - The dynamic behavior of solar prominences and filaments is a precursor 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 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 hybrid algorithmic-data driven method which performance was shown on a set of Solar Dynamic Observatory (SDO) Atmospheric Imaging Assembly (AIA) instrument data performing segmentation of extreme ultraviolet (EUV) and shock waves. In this work, we extend certain aspects of this approach to image segmentation of solar features -- filaments and prominences using H-alpha Kanzelhohe Observatory data. We describe a leakage-aware, sparse multi-year sampling strategy. We present image segmentation models pre-trained on synoptic and normalized image data, followed by a discussion of how model performance depends on training-set engineering, data preprocessing techniques, and the choice of loss metrics. We perform a limited study, both analytically and numerically, of the effect of external, fixed normalization versus native input for compact U-Net-based segmentation of solar H-alpha filaments, also contributing to ongoing research on normalization in deep networks in general and compact models in particular.
ER -