Fig. 4

Flowchart summarizing the steps currently involved in the making of a physics-based secular variation candidate model. X(t), Y(t), Z(t) are the three components of the magnetic field vector recorded in an observatory or by a satellite. Field modelling refers to the process that converts databases of these time series into global mathematical models of the main magnetic field \({{\mathbf {B}}}(t)\) and its rate of change \({\dot{{{\mathbf {B}}}}}(t)\). These mathematical models are fed, together with their uncertainty, to a workflow that initializes the state vector (or ensemble of state vectors) for subsequent integration by the numerical model. The average secular variation between the start date of the IGRF (\(t_0\)) and its end (\(t_0\) + \(\Delta t=5\) yr) is next readily evaluated. If an ensemble approach is used, an ensemble of such average secular variations are obtained, the statistics of which can be used for uncertainty quantification