Orbit-localised thermosphere density prediction using a Kalman filter based calibration of empirical models

TL;DR

This paper introduces a Kalman filter-based calibration method for empirical thermosphere density models, significantly improving their accuracy in satellite orbit prediction by combining multiple models' estimates.

Contribution

A novel Kalman filter approach for calibrating thermosphere models along satellite trajectories, enhancing density estimation accuracy over traditional methods.

Findings

Significant reduction in density residuals with Kalman filter calibration.

Improved density estimates by combining multiple models using a best linear unbiased estimator.

Enhanced satellite orbit prediction accuracy through calibrated models.

Abstract

Accurate estimation of thermosphere mass density is critical to determining how satellite orbits evolve over time and thus to planning and managing space missions. Empirical thermosphere models are commonly employed for this purpose, but have substantial uncertainties. In this work, a Kalman filter method for calibrating these models along a particular satellite's trajectory is described. This method was applied to calibrate the NRLMSISE-00, JB2008, and DTM-2020 models with respect to densities measured by the Swarm-C satellite. Substantial improvements in root mean squared density residuals were obtained using the technique when compared with either uncalibrated model output or calibration using a linear regression on previous data. Further improvement was obtained by combining estimates from different models using a best linear unbiased estimator method.