Assimilation frameworks for merging accelerometer derived thermospheric neutral mass density estimates with empirical and physical models

Atmospheric drag has a direct relationship with the thermospheric neutral density (TND) and represents a considerable impact on the precise orbit determination (POD) and prediction of low Earth orbit (LEO) satellites, for example those with the altitude of less than 1000 km, as well as space debris. The distribution of TND combined with the solar and geomagnetic activity has a direct impact on the electron distribution in the ionosphere. The latter is important for medium- and long-range high frequency communication, positioning, and over-the-horizon radar systems. New capabilities to understand, model and predict thermosphere variables are typically provided by models; however, the quality of them is limited due to their imperfect structure and uncertainty of their inputs. In this study, we present various data assimilation frameworks to take advantage of freely available accelerometer derived TNDs from GRACE, GRACE-FO and Swarm missions. This is realized by (1) formulating ensemble Kalman filter (EnKF)-based calibration and data assimilation (C/DA) procedures to update the model's states (and simultaneously calibrates its key parameters if needed); and (2) an empirical decomposition-based data assimilation is applied to merge satellite derived along-track estimates with global model derived TND simulations. The results of these two frameworks are then evaluated against independent measurements. The technical challenges and benefits are discussed in detail.