Rachit Bhatia, West Virginia University; Gerardo Rivera, West Virginia University; Jacob Griesbach, ARKA/Stratagem; Piyush Mehta, West Virginia University
Keywords: covariance realism, space weather, optimal sequential filter, space traffic management, orbit prediction
Abstract:
The need to include uncertainty information for probabilistic space weather models, like CHAMP-ML and its drivers, for example, F10.7, is critical for accurate orbit prediction and safe space operation in low Earth orbit (LEO). Under support from the Intelligence Advanced Research Projects Activity (IARPA) Space Debris Identification and Tracking (SINTRA) program and the Office of Space Commerce (OSC), we are developing the next generation drag modeling framework that accurately characterizes atmospheric density uncertainty due to space weather in a physics- and data-driven approach.
This paper introduces a mathematical formulation of the new 1D stochastic unscented transform (SUT) designed to capture the joint statistics of probabilistic models along with probabilistic model drivers. Unlike the traditional (deterministic) unscented transform (UT), the SUT accurately captures the probabilistic characteristics of the model conveyed by the model’s uncertainty output. This capability enables a more accurate modeling of inherently dynamic systems, providing a more realistic representation of the complex dynamics. Due to the improved state estimation and prediction, this framework can find application across diverse fields, proving particularly beneficial in complex environments where deterministic models fall short, and Monte Carlo simulations become inefficient. Notably, the SUT is well-suited for problems with limited computational resources and is suitable for real-time applications. These attributes make it particularly well-suited for use in space missions.
The developed SUT formulation is validated using simple numerical examples of linear and non-linear systems and then applied to the case of drag modeling by incorporating the effects of uncertainty in the solar driver and density models in real-time orbit propagation. Additionally, since probabilistic density models are a recent development, we also develop a new algorithm to incorporate the joint density (dynamic) uncertainty obtained with SUT in orbit propagation. This work moves us in the direction of realistic covariance for conjunction assessment, space traffic management, and space safety and sustainability.
Date of Conference: September 17-20, 2024
Track: Conjunction/RPO