Investigation of the global climatologic performance of ionospheric models utilizing in-situ Swarm satellite electron density measurements

The Swarm constellation is a triplet of satellites, flying, since their final configuration reached in April 2014, at altitudes of about 420 to 490 km (the lower pair) and about 500 to 530 km (the upper satellite). All the three satellites provide in-situ measurements of the plasma density in the topside ionosphere using Langmuir Probe sensors onboard the Electrical Field Instrument. The present study is a comprehensive investigation into the climatologic performance of three ionospheric models when compared to the Swarm satellite in-situ measurements. The models are the International Reference Ionosphere (IRI) model, a quick run ionospheric electron density model (NeQuick), and a 3-dimensional electron density model based on artificial neural network training of COSMIC (Constellation Observing System for Meteorology, Ionosphere, and Climate) satellites radio occultation measurements (3D-NN). The mean monthly quiet-time latitudinal profile of Swarm measurements was computed by binning the Swarm electron density measurements in 15-degree longitudes starting from longitude −180° in steps of 15° to 180°, and corresponding model predictions were obtained. The data used in the study covers the years 2014, 2016, 2019, and 2022, capturing various phases of the solar activity cycle. Results from the study show that modelled electron density predictions from all three climatologic models are fairly good representations of the Swarm satellite measurements, with some exceptions in which the models underestimate or overestimate the Swarm satellite values. The IRI model performed best at the northern hemisphere mid latitude, and it overestimated the Swarm measurements at altitudes of ∼ 450 km, especially at the southern hemisphere mid and high latitudes. The NeQuick performed best during the night times, and it overestimated the Swarm measurements, especially at the mid latitudes. The NeQuick was also observed to overestimate the Swarm measurements during the winter solstices at both hemispheres, which is June solstice in the southern hemisphere and December solstice in the northern hemisphere. Overall, the 3D-NN model most often performed better than the IRI model and the NeQuick, especially during the day times and during the high solar activity year (2014), but it underestimated the Swarm measurements, especially at the low and mid latitudes. For all categories explored in the study, the 3D-NN consistently performed better than the other two models. The NeQuick performed better than the IRI model at altitude of satellite B, while the IRI model performed slightly better than the NeQuick at altitude of satellites A and C. The NeQuick also performed better than the IRI model in the local time category, whereas the IRI model performed better than the NeQuick in categories of season, solar activity, and longitudinal sector.