The Earth’s magnetosphere shields our planet from hazardous space weather effects caused by solar disturbances and energetic particles. For this reason, it is critically important to know the global structure of this shield as well as when, where, and how external disturbances can potentially penetrate it. However, the description of the global structure of the magnetosphere and its changes during major disturbances, such as storms and substorms, is extremely difficult, because the magnetosphere is very sparsely sampled by in situ observations. At any moment there are fewer than a dozen dedicated probes beyond low Earth orbit (LEO). One way to mitigate this problem is to probe the magnetosphere remotely – from space, using X-ray and energetic neutral atom (ENA) emissions resulting from charge exchange between plasma and neutral gas, from constellations of LEO spacecraft (e.g., ~70 Iridium® probes providing distributed measurements of Birkeland field-aligned currents), from ground radars and all-sky cameras, or space-borne auroral imagers. A complementary approach is to employ global first-principle models to describe the magnetospheric structure and evolution, linking solar wind perturbations to their ultimate space weather impacts. Finally, historical databases of in situ spacecraft observations (e.g., the magnetic field) or extended LEO observations of near-Earth regions (e.g., the plasmasphere) can be mined to build sophisticated empirical models (e.g., artificial neural networks) that learn from data and improve as the data volume increases (a distinctive feature of machine learning). All of these approaches are now sufficiently mature, but remain limited in their capabilities if taken alone. It is, therefore, particularly timely to establish an ISSI team with a balanced mix of experts in the corresponding areas to attack the problem of the description of the magnetosphere as a global system in a concerted fashion. We will concentrate on the following unanswered questions:

  1. What are the global storm/substorm distributions and variations of magnetospheric electric currents, plasma pressure and density, and how well can they be reproduced with models and remote sensing, given the extreme scarcity of in-situ observations?
  2. What are the feasible and appropriate methodologies for combining and constraining of global simulations with empirical reconstructions and with global images of the magnetosphere, obtained with ENA, X-ray, low-altitude or ground-based  measurements?

Answering these questions will lead to a major advance in understanding of the global magnetosphere and key space weather perturbations. It will guide future constellation-class missions by providing their virtual proxies built from historical data using data-mining and machine-learning methods. It will advance global first-principles simulations of the magnetosphere by identifying key regions where the models fail to reproduce observed global structures and by adjusting the models in those regions — either by direct ingestion of data or by identifying key missing processes.