Onboard Lunar Orbit Determination Using One-Way Uplink Doppler data
Description
This dataset supports the study “Onboard Lunar Orbit Determination Using One-Way Uplink Doppler: Effects of Orbit Geometry and Receiver-Clock Stability under Force-Model Mismatch.” It contains the processed simulation data used to reproduce all nine figures and ten tables in the manuscript. The data were generated from Monte Carlo simulations of a spacecraft in low lunar orbit tracked intermittently by three Chinese Deep Space Network stations using one-way uplink Doppler. Doppler measurements were processed onboard using an eight-state square-root unscented Kalman filter that estimated spacecraft position, velocity, receiver-clock drift, and the solar-radiation-pressure coefficient. The simulations examined four orbit geometries, three receiver-clock stability levels, station-frequency residuals, and station time-tag residuals under a reduced onboard force model. Files beginning with figure_ contain the numerical data underlying the correspondingly numbered manuscript figures. These include force-acceleration histories, time-dependent median and 90th-percentile position errors, and per-realization comparisons between empirical position error and formal filter uncertainty. Files beginning with table_ contain the scenario definition, test matrix, accuracy statistics, and filter-consistency statistics reported in the correspondingly numbered manuscript tables. All Monte Carlo cases contain 108 realizations. Column names specify the represented quantity and its unit. Position and velocity accuracy statistics are evaluated over the final 24 hours of each three-day simulation unless otherwise stated. The data show that one-way uplink Doppler can support onboard lunar orbit determination, although performance depends on orbit geometry, receiver-clock stability, ground-reference errors, force-model uncertainty, and the observability of estimated force parameters.
Files
Steps to reproduce
The data were produced by numerical simulation. No physical instruments or experimental samples were used. Each simulation covered a 72 h lunar-orbit arc beginning on 8 January 2022 at 00:00 UTC. A spacecraft was tracked by the Kashi, Jiamusi, and Zapala stations using simulated one-way uplink Doppler at 7.17 GHz. Each observation was formed as a 10 s Doppler count. Station access required a minimum elevation of 10 degrees and an unobstructed station-to-spacecraft line of sight. Twenty-four 60 min tracking windows were scheduled, with at least 120 min between windows. Near-perilune access was prioritized for eccentric orbits because the rapid velocity change provides a stronger Doppler signature. Truth trajectories were propagated in a Moon-centred J2000.0 frame using the DOP853 numerical integrator with relative and absolute tolerances of (10^{-11}). Truth dynamics included the GRGM660PRIM lunar gravity field through degree and order 420, Sun and Earth point-mass gravity using the JPL DE441 ephemeris, cannonball solar radiation pressure with (C_R=1.5), and relativistic accelerations associated with the Sun and Jupiter barycentre. The onboard model used the independent AIUB-GRL350A lunar gravity field through degree and order 100. Its solar-radiation-pressure coefficient began at 1.0 and was estimated by the filter. Doppler counts were processed with an eight-state square-root unscented Kalman filter. The estimated state contained spacecraft position, velocity, receiver-clock drift, and the solar-radiation-pressure coefficient correction. State-noise compensation was applied to prevent the orbit covariance from collapsing when using the reduced onboard gravity model. The simulations were implemented in Python using the Small-Body and Planets Orbit Determination Toolkit (SPOT), SPICE kernels for reference frames and ephemerides, and SciPy numerical integration. Each simulation cell contained 108 Monte Carlo realizations. Initial position, velocity, and receiver-clock errors were drawn independently from zero-mean Gaussian distributions with one-sigma component scales of 1 km, 1 m/s, and (10^{-12}), respectively. The physical truth value (C_R=1.5) was fixed across realizations. Controlled comparisons examined four lunar-orbit geometries, three receiver-clock stability levels, three station-frequency residual amplitudes, and three station time-tag residual amplitudes. Position and velocity errors were calculated against the known truth trajectory. Per-realization accuracy statistics were evaluated over the final 24 h. The reported median and 90th percentile were calculated across the 108 realizations. Normalized Innovation Squared (NIS), Normalized Estimation Error Squared (NEES), and empirical-to-formal uncertainty ratios were calculated from the same final-24-h interval.
Institutions
- Wuhan UniversityHubei, Wuhan