Modern radiation belt models can be broadly split into either physics-driven diffusion based algorithms, or data science techniques that utilise the continuous data coverage from satellites at geostationary orbit and the Lagrange point L1 to apply statistical data analysis methods to predict electron fluxes at geostationary orbit. The first kind, while posessing generality due to their physical nature, lack accuracy compared to their data-driven counterparts. This is because the magnetosphere is a highly complex system that is not easy to model and its dynamics are not yet fully understood. Meanwhile, data-driven methods possess statistical accuracy, but cannot predict outside their operating parameters, and so on their own provide no information about what happens in the wider radiation belt region.
This thesis is devoted to the development of a model that combines the two approaches into one unified system, attempting to combine the predictive range of physical modelling with the accuracy of the data-driven approach. This model uses geosynchronous orbit fluxes predicted using an advanced data science technique as an input to drive a physics-based radiation belt modelling code. The model has been developed and tested for a range of energy channels, magnetospheric conditions, and with various modifications. It was validated using data from NASA's recent Van Allen probes mission and with NOAA's GOES-13 geostationary satellite. The model results are in good agreement with observations, with the sources of inaccuracies explored in the manuscript. This work is a first attempt to create such a model, and potential improvements are outlined that should further increase accuracy.
A further modification of the model is explored that is found to provide superior performance at geostationary orbit at the cost of degraded performance elsewhere. It is proposed to use this modification in tandem with the main model, where accurate information about geostationary orbit is required. The modification has been tested on long-duration time periods and was found to generate good predictions for high-energy electron fluxes.
The role of electromagnetic ion cyclotron (EMIC) waves is explored using wave vector analysis and calculation of minimum resonant energies. The aim is to identify what effect EMIC waves have on electron dynamics at energies below 1 MeV. The conclusions are that EMIC waves, under certain conditions, do affect these electron populations in the magnetosphere, and their effect should be included in a representative radiation belt model. This is suggested as a further improvement to the simulation.
