Statistical Models of Ionospheric Variability and Irregularities in the Topside Ionosphere

The Earth’s ionosphere can be driven by the Sun, the solar wind, the magnetosphere, as well as the neutral atmosphere. These drivers influence the ionosphere on a variety of spatial and temporal scales. The ionosphere is highly dependent on the driving processes and is highly dynamic. Modelling this plasma and capturing its full dynamic range is challenging.

Swarm is the European Space Agency’s (ESA) first constellation mission for Earth Observation (EO), comprising multiple satellites in Low Earth Orbit (LEO). Numerous data products are available, including measures of the ionosphere at a range of spatial scales. During the Swarm-VIP-Dynamic project, which ended in February 2026, the technique of Generalised Linear Modelling was used to create a suite of statistical models. These models predict the electron density and the variability in the ionospheric plasma at spatial scales between 100 km and 7.5 km. The models were based upon proxies for the heliogeophysical processes, as well as measurements of the thermosphere and ionospheric current systems. In addition to the Swarm data, datasets from other satellites and ground-based instruments were used for model evaluation and validation activities.

The performance of the models of the electron density approached the theoretical best values for some of the goodness-of-fit statistics that were to evaluate these models. This suggests that the modelling method is appropriate for the task undertaken. The models of ionospheric variability at larger spatial scales (~100 km) also performed well, however the model performance decreased at smaller spatial scales. This suggested that there is a physical process missing from the models. Possible candidates are instability processes or driving of the ionosphere by wave activity from below, neither of which are captured by the models at present. It is possible to test whether atmospheric waves originating in the lower atmosphere are driving the variability at European midlatitudes using different proxies for wave activity, and the ways in which this could be tested are discussed.