A Two‐Step Method for Enhancing Thermospheric Mass Density Empirical Model

Abstract With the rapid development of Low Earth Orbit constellations, the importance of accurately estimating thermospheric mass density is steadily increasing. Current empirical models still exhibit unavoidable biases in estimating thermospheric mass density. In this paper, we propose a novel two‐step empirical model calibration enhancement method. The enhancement coefficients are constructed by the second‐order spherical harmonic function and the machine learning method. The daily in situ mass density observations collected by the SWARM‐C and GRACE‐A satellites between 2015 and 2017 are used to generate coefficients and test the model. The two‐step enhancement method performs well under various conditions, including different satellites, altitudes, and space weather scenarios. During storm periods, using satellite observations as the references, the enhanced MSIS 2.1 model can achieve estimation error reduction of over 15%. During long‐term testing in 2015, 2016, and 2017, the average RMSEs of the enhanced thermospheric mass density decreased by 23.06%, 20.33%, and 30.95%, respectively.