As the Sun Turns: Long-term probabilistic forecasting of the GEO electron environment.

Traditionally, models of the electron environment have focused on deterministic forecasts with horizons of hours or days. Here, we take a different approach and present a new set of models that exploit the repeatability of the solar rotation to give long term probabilistic forecasts. The models employ a machine learning technique called random forest regressors. Using inputs of the time history of fluxes, geomagnetic indices and the solar wind speed we are able to forecast the GEO electron environment for horizons of up to 28 days. We present initial results from more than 1 year of running the model in real-time, as well as validation from several years of out-of-sample test data.