Height determination based on GNSS measurements in the mountainous area: contribution of the geoid model and data processing technique to the overall error budget

Heights provided by GNSS are affected by the quality of the geoid or quasigeoid model used for the transformation of the ellipsoidal heights to the orthometric or normal heights, as well as by the data processing techniques, including Real-Time Kinematic (RTK), Real-Time Network (RTN), static relative baseline, and absolute Precise Point Positioning (PPP) solutions. We employ two geoid models for the Tatra Mountains with constant and variable density of the lithosphere. We compare heights for 113 mountain peaks and passes directly measured using GNSS, applying two geoid models and two quasigeoid models – one dedicated to the Tatra Mountains and the second that is used as a national standard for GNSS applications in Poland. We also compare the results of height determination based on static GNSS measurements and post-processing to those based on RTK, RTN, and PPP. We found that the maximum differences from using different geoid and quasigeoid models reach up to 7.5 and 11.6 cm, respectively, whereas the standard deviations from height differences based on different GNSS processing techniques are just 0.8 cm with a maximum difference of 2.4 cm. Wrong tropospheric delay handling may result in an error of 17 cm. Hence, the geoid and quasigeoid models are crucial in GNSS height determination of the mountain peaks, whereas the GNSS data processing technique plays a minor role. Therefore, quick real-time RTN solutions are fully applicable for the GNSS measurements of mountain peaks and passes, even if the height difference between the reference station and the rover exceeds 1800 m, provided that the tropospheric delay is properly corrected by extrapolation or estimation.