Prediction of the ionospheric parameters

Introduction to the methods

Rapid GIM

We implemented daily deep ensemble neural networks (NN) to model GNSS vertical electron content (VTEC) and to estimate the associated uncertainty. The predicted VTEC values and their uncertainties are delivered as global ionospheric maps (GIMs), with a spatial resolution of 2.5°x5° and a temporal resolution of 15 minutes. The products are generated with a latency of approximately one day.

 

For user convenience, the products follow the IONosphere Map EXchange (IONEX) format, consistent with those provided by the International GNSS Service. The NN-based GIMs are uploaded to this webserver and are available for download under Products. If you require higher temporal or spatial resolution products, please feel free to contact us. If you use our data, please cite https://doi.org/10.1029/2025SW004446.

Refined real-time GIM

To improve the accuracy of classical real-time global ionospheric maps (GIMs) we implemented a machine learning approach to refine these GIMs to be more consistent with the IGS final GIMs. We operationally download the International GNSS Service (IGS) combined real-time GIM and refine it with our machine learning model. The refined GIMs are uploaded on this webserver and available for download at Products. If you use our data please cite https://doi.org/10.5194/egusphere-egu24-3117.

 

Acknowledgement

For the creation of our product we use data from following sources:

IGS real-time GIMs from here

IGS final GIMs from here