Improving the fixed solution by processing the unmodeled errors in GNSS RTK long baseline positioning

Unmodeled errors materially affect the float solutions of both ambiguities and coordinates in global navigation satellite systems (GNSS), and thus also the fixed solution. Recently, extensive attempts have been made to guarantee the reliability of integer ambiguity or reject observation faults for improving the fixed solution in real-time kinematic positioning (RTK), while far too little studies have achieved this by processing the unmodeled errors in long baselines. This contribution is therefore set out to address the unmodeled errors in GNSS RTK long baseline fixed solution. At first, we establish an equation using the quadratic form of the float ambiguity to estimate the float ambiguity bias. Then, based on the estimated float ambiguity bias, a procedure for processing the unmodeled errors in the fixed solution is proposed. Finally, a simulation experiment and a real-measured experiment are conducted, respectively. By applying the proposed procedure, the improvements in the mean and RMSE of the fixed solution bias are found to be approximately 1.11–6.17 cm and 1.06–5.73 cm in three dimensions.