Kp forecast based on solar wind observations

Description: This forecast is produced at GFZ using real time measurements of solar wind and interplanetary magnetic field. A neural network is trained to predict Kp values using historical data. The trained network is then fed with real time measurements and produces a forecast up to 72 hours ahead.

Ensemble Kp forecast based on PAGER ensemble solar wind predictions

Description: This forecast product is prepared by our Section at GFZ in the context of PAGER HORIZON 2020 project. A machine learning model is trained to predict Kp values using historical solar wind and interplanetary magnetic field data from OMNIWeb. The trained model is then fed with the ensemble forecast of solar wind at L1 produced by the SWIFT module and provides an ensemble forecast of Kp up to 72 hours ahead. In the figure below the bars indicate the ensemble median values for the Kp index, while the red lines indicate the maximum value among the ensemble members for each forecast horizon.

Niemegk observatory (GFZ) Kp Nowcast

This is a reproduction of the Kp index Nowcast produced at the Niemegk geomagnetic observatory (GFZ). The original data can be found here.

Historical comparison between KP Niemegk and Kp 0 h-horizon forecast

Description: We provide here an historical, last two days comparison between the Kp Niemegk and the 0-h horizon forecast, based on a machine learning model fed with real time solar wind measurements from ACE mission and the SWIFT ensemble solar-wind forecast.

Historical comparison between KP Niemegk and Kp 24 h-horizon forecast

Description: We provide here an historical, last two days comparison between the Kp Niemegk and the 24-h horizon forecast, based on a machine learning model fed with real time solar wind measurements from ACE mission and the SWIFT ensemble solar-wind forecast.

SWPC Kp forecast

Description: This is reproduction of the NOAA SWPC Kp forecast. The original data can be found here.