Comparison of GPS-derived ocean tide loading displacements with models in New Zealand

We investigate the tidal displacement signals of two principle semidiurnal (M-2) and diurnal (O-1) tidal constituents in New Zealand. We calculate the predicted loading displacements using the latest ocean tide models and Earth models. The results show that, for the M-2 constituent, the tide models FES2014b and TPXO9-Atlas have subtle differences, and the vector differences between the Earth models shows that the regularity of the influence of the Earth model is weak. The vector difference of O-1 constituent have significant spatial coherence (about 0.2 mm). The residuals of M-2 constituent in the up component subtract about 0.2 mm across the entire areas after considering the anelastic dispersion. We find that the dear regional-scale spatial coherence of the residuals exists in both constituents in all models even if spatial water density and compressibility have been considered. However, the spatial coherence of residuals disappears when the network-mean value removed, and the reduction of RMS for O-1 constituent is close to the theoretical error of body tide (about 0.5 mm). We also find that for the different ocean tide models, part of the differences in residuals between the observations and predictions would be attributed to the choke of geocentric reference frame.