Mathematical Modeling of Thermospheric Mass Density Via Empirical Orthogonal Function Analysis

The accurate prediction of thermospheric mass density is always important because it gives not only deeper insight into the upper atmosphere dynamics but also critical information about satellite drags. In this paper, we build a regression model of thermospheric mass density at 400 km altitude using the Swarm mission (Swarm-C) accelerometer observations during 2014–2021. Using the empirical orthogonal function (EOF) analysis method, thermospheric density data are decomposed into spatial modes and temporal evolutions. An empirical model is developed based on the first three eigenmodes. To study further the factors and mechanisms controlling the thermospheric density variations, we investigate the empirical relation of thermospheric density to latitude, longitude, local time, season, as well as solar and geomagnetic activities. Our empirical model shows a thermospheric equatorial anomaly, hemispheric asymmetry, as well as significant seasonal and local-time dependence of the mass density. For model validation, we compared our model outputs with MSIS data and found a better agreement with Swarm data. Our EOF model can reproduce the global thermospheric mass density and its seasonal variations with the first three dominant modes.