A Realistic and Easy-to-Implement Weighting Model for GPS Phase Observations

Abstract

Observation weighting is an essential component of GPS stochastic modeling and plays a key role in reliable outlier detection and parameter estimation. Nowadays, satellite elevation angle and SNR are used as quality indicators for GPS phase measurements in high-accuracy geodetic applications. In comparison with elevation-dependent models, SNR-based weighting schemes represent the reality better, but usually require greater implementation efforts. Relying upon a representative analysis of empirical SNR-based weights, this paper proposes the elevation-dependent exponential weighting function EXPZ, which benefits from realistic SNR-based weights and enables easy software implementation. To process GPS data from a regional network, this advanced weighting scheme is implemented in the Bernese GPS Software 5.0 and is compared with the conventional elevation-dependent COSZ model in terms of phase ambiguity resolution, troposphere parameter (TRP) estimation, and site coordinate determination. The results show that the proposed EXPZ model significantly attenuates the downweighting effects on low-elevation observations and improves the success rates of ambiguity resolution by about 10%, the standard deviations of site-specific TRPs by about 40%, and the repeatability of daily coordinate estimates by up to 2.3 mm (50%).