From macro models to generic acceleration tables: modeling non-gravitational forces acting on Galileo satellites

Galileo satellite surface properties are published by the European GNSS Service Center (GSC) as part of the official satellite metadata. These properties describe surface elements that are grouped into four main categories: multi-layer insulation (MLI), optical radiators, navigation antennas, and solar cells. Based on this information, a simple box-wing solar radiation pressure (SRP) model is widely applied by the International GNSS Service (IGS) analysis centers (ACs) for Precise Orbit Determination (POD).

However, key thermal effects, such as radiator emission and imbalanced thermal radiation from navigation antennas and solar panels, are typically neglected, as detailed thermal information is not publicly available. Instead, these effects are partially absorbed by the Empirical CODE Orbit Model (ECOM/ECOM2) parameters, which are widely estimated to compensate for deficiencies in the physical force modeling.

In this contribution, we develop advanced macro models for Galileo satellites, including refined SRP and Earth radiation pressure (ERP) models based on a more detailed representation of satellite surface elements, as well as thermal radiation models for radiator emission, navigation antennas, and solar panels using best-guess values. The impact of each individual thermal force component on POD and terrestrial reference frame solutions is assessed. The final results are compared against the standard GSC-based box-wing model.

Given the complexity of the physical macro models, we introduce acceleration tables for Galileo In-Orbit Validation (IOV) and Full Operational Capability (FOC) satellites as generic interface between macro models and the orbit integrator. The tables provide non-gravitational accelerations as functions of satellite orbital argument and Sun elevation angle above orbital plane, including both SRP and thermal radiation effects. Earth radiation pressure is excluded from the tables due to its pronounced temporal variability and should therefore be modeled separately.