How can space-borne along-track neutral density measurements be used to predict multi-level global thermospheric neutral density fields?

An accurate estimation of the Thermospheric Neutral Density (TND) is important for predicting the orbit of satellites and objects, for example, those with the altitude of less than 1000 km. Models are often used to simulate TNDs but their accuracy is limited due to modelling restrictions and sensitivity to the calibration period. Satellite missions such as CHAMP, GRACE, GOCE, Swarm, and GRACE-FO are equipped with on-board accelerometer sensors to measure drag forces, which can be used to estimate along-track TNDs. However, spatial and temporal coverage of these space borne TNDs is restricted to the mission design. To make the best use of the modelling tools and measurements, we applied these along-track TND measurements within the sequential Calibration and Data Assimilation (C/DA) framework proposed by (Forootan et al., 2022, doi:10.1038/s41598-022-05952-y). The C/DA is used to re-calibrate the NRLMSISE00 model, which is called “C/DA-NRLMSISE00”, whose outputs fit well to the introduced space-borne TNDs. The C/DA-NRLMSISE00 is applicable for forecasting TNDs and individual neutral mass compositions at any predefined vertical level (between ~100 and ~600 km) with user-defined spatial-temporal sampling. Nine time periods (October 2003, July 2004, March 2008, April 2010, March 2015, September 2017, August 2018, September 2020 and October 2021) associated with space weather storms are selected for our investigations because most of the available models lack accuracy to provide reasonable TND simulations. Independent comparisons are performed with the space-borne TNDs that were not used within the C/DA framework, as well as with the outputs of other thermospheric models such as Jacchia-Bowman 2008 (JB2008) and the High Accuracy Satellite Drag Model (HASDM) database. The numerical results indicate improvements in the Root Mean Squared Errors (RMSE) of the C/DA-NRLMSISE00's TND forecasts compared to NRLMSISE-00, JB2008 and HASDM along-track of the LEO missions. The percentage reductions are found to be: 51%, 8% and 8 % along GRACE (2003, average altitude 490 km), 25%, 20% and 48% along GOCE (2010, average altitude 270 km), 46%, 37% and 35% along Swarm B (2015, average altitude 520 km), 54%, 12% and 5 % along Swarm B (2017, average altitude 514 km), and 41% and 64% along GRACE (FO) (2021, average altitude 504 km), respectively