Abstract: Multi-epoch geodetic observation datasets from seabed monitoring stations generate coordinate time series, providing crucial support for analyzing crustal deformation and establishing seabed geodetic monitoring networks. Traditional center-point methods, which rely on the centroid datum, are prone to interference from station observation errors. This interference induces potential offsets in the centroid datum reference network, thereby degrading overall computational accuracy. To address this issue, we propose a coordinate time series analysis model based on the weighted centroid datum. Incorporating prior information to construct weighting factors, we establish a weighted centroid datum reference network that better aligns with objective reality. Measured data from the vicinity of the Japan Trench and the South China Sea Trough are used for verification. The results indicate that the weighted centroid datum can improve the fitting accuracy of coordinate time series, its standard deviation does not exceed 0.117 m and the root mean square error of residuals decreases from 0.1226 m to 0.1141 m. Compared with traditional methods, the proposed model demonstrates higher accuracy under multi-epoch observation conditions, while more effectively correcting interference from abnormal stations.