Prediction of plasmaspheric hiss spectral classes

We present a random forests machine learning model for prediction of plasmaspheric hiss spectral classes from the Van Allen Probes dataset. The provides accurate obtained by self organizing map (SOM) unsupervised classification technique. high predictive skill is largely determined distinct and different locations given class (“no hiss”, “regular “low-frequency hiss”) in (MLAT, MLT, L) coordinate space, which are main predictors simplest most base model. Adding to such any other single predictor among magnetospheric, geomagnetic, solar wind conditions only minor similarly incremental improvements skill, comparable one when including all possible predictors, thus confirming major role spatial location prediction.