Copyright © <2024> The Regents of the University of Michigan

GeoDGP Overview

Accurately predicting the horizontal component of ground magnetic field perturbations is crucial for the risk management of natural hazards caused by space weather events, such as geomagnetically induced currents (GICs). GeoDGP is a data-driven model that probabilistically forecasts local geomagnetic perturbations over the globe at 1-minute cadence and at most 1-hour ahead. The model is trained on 30 years of (1995-2022) NASA/GSFC’s OMNI dataset and SuperMAG ground magnetometer measurements.

Check out our real-time global prediction.

Features

Input

• Solar wind measurements from the first Lagrangian point (L1).
• The location input.

Variable Name	Description
$B_x, B_y, B_z$	Interplanetary Magnetic Field (IMF)
$V_x$	Solar wind velocity
$N_p$	Proton number density
$T$	Plasma temperature
Dst	Disturbance storm time index
$\theta$	Dipole tilt angle
$\lambda$	Geomagnetic latitude
$\phi$	Geomagnetic longitude in Solar Magnetic (SM) coordinates

Output

• The north, east, and horizontal components of geomagnetic perturbations ($dB_H$, $dB_N$, and $dB_E$) with a lead time corresponding to 1 hour plus the solar wind propagation time from L1.

Getting Started

1. Clone the repository

git clone https://github.com/HongfanChen/GeoDGP.git

2. Set up a virtual environment (recommended)

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

Structure

GeoDGP/
├── data/               # this can be downloaded from Deep Blue Data, see data section below
├── figure/             # this can be downloaded from Deep Blue Data, see data section below
├── docs/               # documents
├── scripts/            # scripts to reproduce results in the paper
├── src/                # Source code
├── requirements.txt    # Python dependencies
└── README.md

Data Access

To reproduce the results in the original paper, please refer to the dataset available at the Deep Blue Data repository, and find the permanantly archived release of code at .

License

This project is licensed under the PolyForm Noncommercial License 1.0.0 License - see the LICENSE file for details. If you are interested in using the code for commercial purposes, please reach out to Hongfan Chen (chenhf@umich.edu).

Citation

Chen, H., Toth, G., Chen, Y., Zou, S., Huang, Z., & Huan, X. (2025). GeoDGP: One-hour ahead global probabilistic geomagnetic perturbation forecasting using deep Gaussian process. Space Weather, 23, e2024SW004301. https://doi.org/10.1029/2024SW004301

Chen, H., Chen, Y., Huang, Z., Zou, S., Huan, X., Toth, G. Data and Analysis for Global Probabilistic Geomagnetic Perturbation Forecasting Using the Data-Driven Model GeoDGP [Data set], University of Michigan - Deep Blue Data. https://doi.org/10.7302/6brp-0y03

Credits

Hongfan Chen: Conceptualization, Data Curation, Methodology, Formal Analysis, Software.

Gabor Toth: Conceptualization, Supervision, Project Administration.

Yang Chen: Methodology, Supervision.

Shasha Zou: Formal Analysis, Supervision.

Zhenguang Huang: Methodology.

Xun Huan: Methodology, Supervision.

Aniket Jivani: Website Development.

Contacts

Please contact Hongfan Chen (chenhf@umich.edu) for any questions.

Acknowledgement

This work is supported by the National Science Foundation (NSF) under Grant No. 2027555. NextGen Space Weather Modeling Framework Using Data, Physics and Uncertainty Quantification. We acknowledge use of NASA/GSFC’s Space Physics Data Facility’s OMNIWeb and CDAWeb service, and OMNI data. We gratefully acknowledge the SuperMAG collaborators. The Dst data are provided by the WDC for Geomagnetism, Kyoto.