Eric Fattig, West Virginia University; Rachit Bhatia, West Virginia University (Consultant); Thomas Kelecy, West Virginia University (Consultant); Piyush Mehta, West Virginia University
https://doi.org/10.64861/OTRO8706
Keywords: Temporal auto-correlation, atmospheric density, covariance realism, orbital mechanics, orbit determination
Abstract:
Atmospheric drag, driven by upper atmospheric density, significantly influences the orbits of Resident Space Objects (RSOs) in low Earth orbit (LEO). Accurate space situational awareness (SSA) and space domain awareness (SDA), including conjunction assessments and RSO tracking, depend on reliable models of atmospheric density variability and its uncertainty. Although progress has been made in understanding the variability drivers of the upper atmosphere, uncertainty in density modeling and characterization of its temporal autocorrelation remain underexplored. This work advances covariance realism in LEO orbit predictions by incorporating atmospheric density uncertainty directly into the state vector using a first-order Gauss-Markov (FOGM) process.
Initially presented at the AMOS 2025 conference through the Dynamic Orbit Propagator and Estimator (DOPE) framework, previous work incorporated atmospheric density uncertainty solely during the propagation phase using simplified force models (i.e., two-body dynamics combined with J2 perturbations and drag). Here, we extend the investigation by integrating simulated observations and a high-fidelity force model into an orbit determination/propagation (OD/OP) architecture. The enhanced DOPE framework now includes a normalized representation of the density-error (delta-rho) in the state vector and allows the incorporation of high-fidelity force modeling (gravitational harmonics up to 96×96, atmospheric drag, third-body effects, Earth solid tides, and solar radiation pressure) to assess sensitivity to the dynamics essential to high accuracy LEO predictions.
The study systematically compares orbit propagation results with and without the inclusion of these additional dynamics, emphasizing the sensitivity of propagation accuracy and covariance realism to the effects of atmospheric drag variability and density autocorrelation. Building upon earlier research that examined the impact of differing density autocorrelation parameter values using unscented transform techniques (validated against Monte Carlo simulations), this work now incorporates simulated observation updates for complete orbit determination, to include the density corrections. Filter performance is evaluated through metrics such as the normalized estimation error squared (NEES) and the normalized innovations squared (NIS). Preliminary validation with simulated observations demonstrates the feasibility of this enhanced DOPE framework for real-time operational applications in space traffic management.
First, the initial DOPE “truth” scenarios are recreated with the addition of the high-fidelity force modeling and compared against the low-fidelity results to ensure consistency and gauge sensitivity of the technique to the additional perturbations. Then, the filter component of the unscented transform is added to the framework (supplementing the existing propagator), such that the normalized density-error is now included in the estimated state. Next, simulated observations data from an on-board Global Navigation Satellite System (GNSS) receiver provides measurement updates which allow complete validation of the OD functionality. Lastly, the OD and prediction performance are evaluated with the metrics mentioned above, allowing conclusions to be drawn on how density autocorrelation parameters impact prediction accuracy and covariance realism.
This work is an important facet in a next-generation atmospheric drag modeling effort supported by the Intelligence Advanced Research Projects Activity (IARPA) Space Debris Identification and Tracking (SINTRA) program and the Office of Space Commerce (OSC). Significant density uncertainty remains a challenge in the context of covariance realism for the purposes of operational awareness in the space domain. The DOPE technique contributes to the progress made by the SSA/SDA community in seeking to improve covariance realism through improvements in density uncertainty modeling in the OD and propagation processes. These results also set the stage for further investigations and development of processes to optimize selection of the density autocorrelation parameter.
Date of Conference: September 15-18, 2026
Track: Astrodynamics