Baseline-dependent performance assessment of an ionosphere-weighted decoupled clock single-difference RTK model

Abstract

Reliable ambiguity resolution and atmospheric residual mitigation remain critical for medium- and long-baseline real-time kinematic (RTK) positioning. Conventional double-differenced (DD) RTK models eliminate receiver clock and hardware biases but amplify observation noise and introduce mathematical correlations, which may degrade ambiguity convergence under baseline-dependent atmospheric decorrelation. In this contribution, an ionosphere-weighted decoupled clock single-difference RTK (DC-SD) model is assessed against a conventional ionosphere-weighted DD model. In the proposed model, receiver-dependent code and phase hardware biases are absorbed into decoupled clock parameters, while stochastic ionospheric pseudo-observations are introduced to constrain between-receiver ionospheric delays and form a full-rank SD model. Then, the observation residual analysis, atmospheric residual assessment, ambiguity dilution of precision (ADOP) analysis, ambiguity convergence evaluation, and ambiguity-converged positioning tests are conducted under a unified processing strategy. After covariance whitening, the two models exhibit similar pseudorange residual characteristics, whereas the DC-SD model shows substantially smaller and more stable normalized carrier-phase residual dispersion, particularly for the longer baselines. After transforming the atmospheric estimates into a common between-satellite differenced domain, the DC-SD model yields smaller ionospheric root mean square (RMS) values over all four baselines, with reductions ranging from approximately 3.5%–23.8%. For the slant tropospheric residuals, the two models show comparable performance on the shorter baselines, while the RMS is reduced by approximately 34.8% and 41.4% for the 43 and 73 km baselines, respectively. The ADOP remains mostly below 0.12 cycles after convergence. Moreover, the mean time to first fix (TTFF) is shortened to 6, 6, 6, and 7 epochs, compared with 34, 39, 79, and 81 epochs for the DD model. The proposed model achieves centimeter-level positioning accuracy with consistently improved horizontal performance. These results demonstrate that the proposed DC-SD model improves residual stability and ambiguity fixing efficiency while maintaining positioning performance comparable to conventional DD RTK.