Abstract
Maintaining accurate real-time hindcast and forecast specification of
the radiation environment is essential for operators to monitor and
mitigate the effects of hazardous radiation on satellite components. The
Radiation Belt Forecasting Model and Framework (RBFMF) provides
real-time forecasts and hindcasts of the electron radiation belt
environment, which are used as inputs for the Satellite Charging
Assessment Tool (SatCAT). We evaluated the long-term statistical error
and bias of the RBFMF by comparing the 10-hour hindcast of electron
phase space densities (PSD) to a multi-mission dataset of PSD
observations. We found that, between the years 2016-2018, the RBFMF
reproduced the radiation belt environment to within a factor of 1.5.
While the error and bias of assimilated observations were found to
influence the error and bias of the hindcast, data assimilation resulted
in more accurate specification of the radiation belt state than
real-time Van Allen Probe observations alone. Furthermore, when real
time Van Allen Probe observations were no longer available, the hindcast
errors increased by an order of magnitude. This highlights two needs;
(i) the development of physics-based modelling incorporated into this
framework, and (ii) the need for real-time observations which span the
entire outer radiation belt.