Nonlinear dynamic systems modeling using Gaussian processes : predicting ionospheric total electron content over South Africa

摘要

Two different implementations of Gaussian process (GP) models are proposed toestimate the vertical total electron content (TEC) from dual frequency Global PositioningSystem (GPS) measurements. The model falseness of GP and neural network modelsare compared using daily GPS TEC data from Sutherland, South Africa, and it is shownthat the proposed GP models exhibit superior model falseness. The GP approach hasseveral advantages over previously developed neural network approaches, whichinclude seamless incorporation of prior knowledge, a theoretically principled method fordetermining the much smaller number of free model parameters, the provision of estimatesof the model uncertainty, and a more intuitive interpretability of the model.