Short-Session GNSS Monitoring of Normal Heights Above Underground Mine Workings: Optimisation of a Multi-Reference Configuration and Inter-Cycle Comparison with Precise Levelling

Reliable monitoring of an underground mining area requires separating movement of the monitored surface from instability of the reference frame while maintaining the feasibility of observations within short field windows. At a monitoring site covering approximately 2.2 km², twelve monitoring benchmarks are located above the surface projections of mapped mine-working corridors, while Burovaya, Poselkovyi, and 5784 form an external triangular reference subnetwork; Burovaya was adopted as the primary point, and the coordinates of the two secondary stations are re-determined in each cycle. Five cycles of dual-frequency global navigation satellite system (GNSS) observations were conducted using the Global Positioning System (GPS), the Global Navigation Satellite System (GLONASS), and Galileo jointly. The same rover records were reprocessed under nine conditions: one, two, and three reference stations with nested intervals of 15, 30, and 60 min. Ellipsoidal heights were converted to normal heights in the Baltic height reference frame using a local transformation with a planar approximation of the height component, without applying a geoid or quasigeoid model. Interferometric synthetic aperture radar (InSAR) data from 21 Sentinel-1C epochs spanning 26 May 2025 to 2 June 2026 were used for the spatio-temporal characterisation of the reference configuration; a relative line-of-sight (LOS) velocity of +0.0 ± 1.9 mm/year (formal 1σ) was obtained for Burovaya; its magnitude was smaller than the quoted formal uncertainty and was not statistically significant. This result supports the relative stability of the station within the sensitivity of the method and the observation period. Precise geometric levelling was performed in Cycles 3–5; external evaluation was conducted at Rp-1…Rp-12. The cycle-specific root mean square difference (RMSE) values for the ‘three reference stations, 60 min’ mode were 4.6, 3.5, and 2.4 mm for Cycles 3, 4, and 5, respectively. For the 3→4 and 4→5 transitions, GNSS and levelling indicated the same direction of change at all 12 benchmarks. Paired analysis of Cycles 3–5 at 60 min confirmed the advantage of the tested three-reference configuration over the one- and two-reference configurations, whereas the advantage of 60 min over 30 min was small and statistically non-significant. Therefore, 60 min is regarded as a practical and robust working interval among those tested, rather than as a statistically demonstrated optimum.