Current empirical and machine-learning models in the group
Our group develops predictive models for electron and ion populations in near-Earth space
Soft protons to support SMILE/SXI
Linear proton-intensity models
Five linear-regression models predict 92.2-159.7 keV proton intensities along the SMILE trajectory using Cluster/RAPID and OMNI data
Soft protons to support SMILE/SXI
XGBoost soft-proton forecasting
A framework uses XGBoost and solar-cycle-informed synthetic datasets to forecast soft-proton fluxes for mission planning
Soft protons to support SMILE/SXI
Ensemble shutter-operation classifiers
Ensemble classifiers are being developed to predict high soft-proton flux periods and optimize SMILE/SXI shutter operations
Electron Environment
GENET
A global empirical model of near-Earth electron fluxes with pitch angle resolution based on Cluster/PEACE&RAPID observations
Radiation Belts
MERLIN
A LightGBM model for medium-energy electron flux in Earth’s outer radiation belt, trained on long-term GPS observations
Polar cap ion outflow
Cold-ion outflow prediction
Machine-learning models, including Extra-Trees regression, predict cold-ion fluxes from Cluster/EFW&EDI&FGM observations, solar activity, and solar wind activity