Eclipse Transient Orbit Determination

Leveraging statistical atmospheric models for autonomous navigation

Summary

  • Eclipse transients as observables for autonomous navigation.
  • UKF estimation with statistical atmosphere modeling.
  • Published AIAA SciTech 2024 results.

Overview

When a satellite crosses into or out of Earth's shadow, it experiences measurable changes in thermal state, power generation, and sensor readings. These eclipse transients contain geometric information about the satellite's orbit—information that can be exploited for autonomous navigation without ground station communication.

This research developed a modular simulation framework for modeling eclipse transients and demonstrated how Unscented Kalman Filtering (UKF) can estimate orbital state from these measurements. A key contribution was incorporating statistical atmospheric models derived from real-world climate data, capturing variability in atmospheric refraction that deterministic models miss.

The Problem

Traditional orbit determination relies on ground-based tracking or GNSS. For missions requiring autonomous navigation—whether due to communication constraints, operational resilience, or cost— alternative observables are needed. Eclipse boundaries provide a geometric constraint: the satellite must lie on a cone defined by the Sun-Earth geometry. By observing when eclipse transitions occur, we can constrain the orbit.

However, Earth's atmosphere complicates matters. Atmospheric refraction bends light near the limb, and density variations cause the effective shadow boundary to fluctuate. Ignoring this uncertainty degrades navigation accuracy significantly.

Technical Approach

Eclipse Driver

Modular Python architecture for simulating eclipse entry/exit with configurable measurement sampling and noise models

Statistical Atmosphere

Climate-derived density profiles capture real-world variability rather than fixed models like US Standard Atmosphere

UKF State Estimation

Unscented Kalman Filter propagates uncertainty through nonlinear eclipse geometry without linearization errors

Sensitivity Analysis

Quantified how atmospheric model uncertainty affects navigation accuracy across different orbital regimes

Key Parameters

Measurement Config Sample period, duration, noise sigma, logistic softness for shadow transition modeling
Filter Config Process/measurement covariance (6×6), UKF tuning parameters (α=0.3, β=2.0, κ=0.0)
Atmospheric Model Statistical density profiles from climate data with atmospheric refraction computation

Key Contributions

  • Modular Python software architecture enabling rapid iteration on eclipse simulation configurations
  • Statistical atmospheric model that captures real-world density variability from climate data
  • UKF-based state estimation framework for eclipse transient measurements
  • Quantitative analysis of filter sensitivity to atmospheric model uncertainty