Quality-Audited Realisation of High-Precision Local GNSS Reference Benchmarks: Implications for Earth-System Monitoring

A defensible local global navigation satellite system (GNSS) reference requires more than a precise network adjustment when small position or displacement changes are interpreted against that reference. This study evaluates a quality-audited realisation workflow for six rooftop control points observed in repeated 1 Hz static GNSS campaigns, with end-to-end documentary and computational traceability from raw observations and processing decisions to reference handling, stochastic modelling, adjustment, and external consistency. Within-session quality was assessed through ambiguity behaviour, residual diagnostics, and network closure, while repeated occupations and reference sensitivities were evaluated separately. After removal of a temporally localised C12 anomaly by a common trim, accepted loop closures ranged from 0.100 to 0.917 mm. Under a nominal 2.8 mm working component scale within a broader data-compatible stochastic family, the posterior variance factor was close to unity, and model-conditional ENU (East, North, Up) standard uncertainties were approximately 2.0–2.4 mm. Component-wise cross-campaign realisation differences were 3.42–6.61 mm after separate translation alignment of the March components, corresponding inter-triangle relative-vector uncertainties were 5.3–5.5 mm, and global positioning system (GPS)-only precise point positioning (PPP) showed centimeter-level external consistency. These results demonstrate that within-session consistency, reoccupation-inclusive assessment, model-conditional uncertainty, datum/reference sensitivity, and external consistency provide complementary, non-interchangeable evidence. The framework provides local reference infrastructure for downstream GNSS-based Earth-system interpretation without claiming direct estimation of Earth-system variables.