Over the past decade, there have been significant advancements in both the availability and diversity of satellite navigation systems, on both regional and global scales. The growing significance of GNSS across various fields has led not only to the modernisation of existing systems like the GPS and GLONASS, but also to the creation of new ones such as Galileo. The Navigation Support Office (NSO) operates as a vital component within the European Space Agency, specialising in the precise utilisation of navigation satellite systems. In 2024, NSO underwent a revision of its GNSS processing system, which resulted in a state-of-the-art GNSS processing framework known as CHAMP, Consolidated High Accuracy Multi-GNSS Processing, which is based on constellation-wise data processing and normal equation stacking to efficiently generate GNSS products for all major navigation constellations. This new approach enables internal comparisons between individual constellation solutions and the consolidated multi-GNSS solution, revealing unique characteristics of each constellation and demonstrating the enhanced impact of combining them in critical ways.

This thesis explores and addresses key inconsistencies in the processing of multi-GNSS solutions within the Navigation Support Office, with a particular focus on producing accurate single-constellation solutions that can be reliably combined into a coherent multi-constellation framework. To tackle these inconsistencies, a comprehensive Python-based programme has been developed as an extension of the CHAMP framework, enabling a more efficient and structured approach for analysing large-scale GNSS data. As a result, this new tool provides the basis needed to identify anomalies, track trends, and address inconsistencies across various constellations during the combination process, as a post-processing analysis or within the operational pipeline. The core of this research involves a detailed examination of the parameters estimated in the NSO GNSS precise orbit determination, both in terms of individual and combined multi-GNSS solutions, including Earth orientation parameters, satellite state vectors, ground station positions, clock biases, and empirical accelerations, among others.

The investigation lead to significant inconsistencies in the Length of Day estimates of the individual GPS constellation, affecting the accuracy of the combined multi-GNSS solution. These issues are compounded by the combination of the large weight given to the GPS constellation in the combination due to its low variability, and the GPS non-gravitational a priori models, specially regarding forces like solar radiation pressure, which have been demonstrated to be not as refined as those for other constellations. Additionally, the tuning of empirical accelerations has shown that there is room for improvement to better reflect satellite dynamics.

In conclusion, several key enhancements were identified and proposed to improve the overall performance and reliability of the combined multi-GNSS solution, and subsequently improve the quality of the NSO GNSS products. These mainly comprise the comparison of different non-gravitational a priori models combined with the fine-tuning of empirical accelerations, allowing for a more robust integration of single-constellation solutions into a cohesive multi-constellation framework. Ultimately, recommendations on the way forward are given to create new models based on the results shown on this thesis