Raw e-callisto data used in classification study ~2020-2023

Raw e-Callisto Data for Type II and Type III Solar Radio Burst Classification NOTE: These files are only provided for the purpose of reproducing the mentioned study. Please consult the e-callisto data archives to ensure attribution is correctly awarded to the e-callisto PI's when using the raw data. Description This dataset contains raw solar radio spectrogram data files used in the study "Type II and Type III Solar Radio Burst Classification Using Transfer Learning" (le Roux et al., 2025, DOI: 10.1007/s11207-025-02595-w). Dataset Contents The dataset is organized into three main categories: 1. Type II Solar Radio Bursts Raw FITS files containing observations of Type II solar radio bursts, characterized by slow-drifting emission patterns typically associated with shock waves generated by coronal mass ejections (CMEs). These events drift from high to low frequencies at a slower rate and appear as sloped structures in radio spectrograms. 2. Type III Solar Radio Bursts Raw FITS files containing observations of Type III solar radio bursts, characterized by fast-drifting emission patterns associated with energetic electrons ejected during solar flares traveling along open magnetic field lines. These events appear as rapidly drifting streaks in spectrograms. 3. Empty/Baseline Category Raw FITS files containing "quiet" spectrograms with no recorded solar radio burst activity. This baseline category consists of stratified samples from January 2020 where no solar radio bursts were reported according to the e-Callisto burst catalogs. Note: This repository contains only the first 7 days of data from January 2020, representing a subset of the full empty category that was sampled from in the original study. Data Source and Coverage Source: e-Callisto network (www.e-callisto.org) Time Period: January, 2020 – April, 2023 Event Labeling and Verification Solar radio burst events were identified and classified using event lists compiled by Christian Monstein, e-Callisto principal investigator and former engineer at IRSOL (Istituto Ricerche Solari Locarno). Event lists are publicly available from the e-Callisto burst catalog: Reference: https://soleil.i4ds.ch/solarradio/data/BurstLists/ Related Materials Processed Dataset and Code:The pre-processed spectrogram images (128×128 pixel grayscale), trained model weights, and complete source code used in the study are available in a companion Zenodo repository: DOI: 10.5281/zenodo.15754977 Published Article: le Roux, H., Steyn, R., Strauss, D.T., Daly, M., Gallagher, P.T., Scully, J., Maloney, S.A., Monstein, C., & Drevin, G. (2025). Type II and Type III Solar Radio Burst Classification Using Transfer Learning. Solar Physics, 300, 179. DOI: 10.1007/s11207-025-02595-w Known Limitations Imbalanced Classes: Type II events (~1,200 samples) are significantly underrepresented compared to Type III events (~16,500 samples) Data Quality: Some samples exhibit varying noise levels due to different station installations and environmental conditions