Forecasting Magnetospheric Dynamics Using Solar Wind Measurements: Role of Global and Multi-Scale Processes

Earth's magnetosphere is a nonlinear spatially extended system which is driven far from the equilibrium by the turbulent solar wind. On the global scales the magnetospheric dynamics appear to be well organized and coherent and therefore allow deterministic considerations. However, on the smaller scales it exhibits scale-invariant behavior which is retained over a wide range of spatial and temporal scales. These multi-scale features do not reveal low effective dimension and therefore can not be predicted dynamically. At the same time, the input-output statistical analysis shows that the scale-free properties of magnetospheric dynamics are strongly affected by the solar wind and therefore can be described in terms of conditional probability. This allows constructing a unified data-derived model of solar wind - magnetosphere coupling which combines the elements of nonlinear dynamical approach with Bayesian consideration of statistical physics. It yields deterministic predictions of the global component of the dynamics and probabilistic predictions of its multi-scale features.