Accuracy assessment of real-time single-receiver GNSS velocimetry in marine dynamic environments

Abstract High-rate Global Navigation Satellite System (GNSS) observations enable real-time single-receiver velocity determination in marine environments. This study systematically evaluates raw Doppler velocity estimation (RDVE), time-differenced pseudorange (TDPR), and time-differenced carrier phase (TDCP) using 5 Hz buoy and shipborne data, broadcast ephemerides, and observations from the Global Positioning System (GPS), BeiDou Navigation Satellite System (BDS), and combined GPS+BDS. A unified processing strategy with consistent stochastic modeling, quality control, and method-specific reference alignment is applied. RDVE is evaluated against the same-epoch reference velocity, whereas TDPR and TDCP are evaluated against the corresponding 0.2 s interval-average reference. For statistical evaluation, finite synchronized estimates are screened using a consistent case-specific residual criterion, with the corresponding finite and retained epoch counts reported separately. In the buoy experiment, GPS+BDS RDVE achieved horizontal and vertical root-mean-square errors (RMSEs) of 0.0160 and 0.0344 m/s, respectively, while GPS+BDS single-frequency TDCP (TDCP-SF) achieved 0.0111 and 0.0237 m/s. In the shipborne experiment, GPS+BDS TDCP using the narrow-lane combination (TDCP-NL) achieved horizontal and vertical RMSEs of 0.0635 and 0.0551 m/s, respectively, with 99.86 % availability. TDCP showed the most stable performance under shipborne dynamics, whereas TDPR was more sensitive to pseudorange noise, unfavorable satellite geometry, and occasional large errors. These results indicate that the practical benefit of multi-GNSS velocimetry depends on observation type, signal availability, and reference-time alignment rather than constellation count alone.