IonoBench Models

IonoBench: Evaluating Spatiotemporal Models for Ionospheric Forecasting under Solar-Balanced and Storm-Aware Conditions
Published in Remote Sensing (MDPI)

Contents

Each model folder contains:

• Best Checkpoint
  Format: MODELNAME_SessionNAME_best_checkpoint_yyyymmdd_hhmm.pth
  Example: SimVP_AllFeatures/model_best.pth

• Training Logs
  Format: MODELNAME_SessionNAME_lrXX_bsXX_yyyymmdd_hhmm.txt
  Example: SimVP_AllFeatures/training_log.txt

• Test Results
  Format: testing_info_yyyy-mm-dd_hh-mm.txt
  Contains evaluation metrics such as RMSE, R², and SSIM on test and storm periods

Notes

• Original configuration files are included and reflect the training settings used.
• A layered configuration structure (base → model → mode → CLI) was adopted later for improved usability.
• The pretrained models are intended for reproducibility and evaluation; training tutorials and CLI tools are available on the GitHub page.

Citation

If you use these models, please cite:

Mert C. Turkmen, Yee Hui Lee, Eng Leong Tan (2025).
IonoBench: Evaluating Spatiotemporal Models for Ionospheric Forecasting under Solar-Balanced and Storm-Aware Conditions.
Remote Sensing, 17(15), 2557. https://doi.org/10.3390/rs17152557

As well as the original refences:

• SimVPv2: Tan et al., 2024 — arXiv:2211.12509
• SwinLSTM: Tang et al., 2023 — arXiv:2308.09891
• DCNN121: Boulch et al., 2018 — arXiv:1810.13273
• OpenSTL: Tan et al., 2023 — arXiv:2306.11249

---