How stationary are the internal tides in a high-resolution global ocean circulation model?

The stationarity of the internal tides generated in a global eddy-resolving ocean circulation model forced by realistic atmospheric fluxes and the luni-solar gravitational potential is explored. The root mean square (RMS) variability in the M2 internal tidal amplitude is approximately 2 mm or less over most of the ocean and exceeds 2 mm in regions with larger internal tidal amplitude. The M2 RMS variability approaches the mean amplitude in weaker tidal areas such as the tropical Pacific and eastern Indian Ocean, but is smaller than the mean amplitude near generation regions. Approximately 60% of the variance in the complex M2 tidal amplitude is due to amplitude-weighted phase variations. Using the RMS tidal amplitude variations normalized by the mean tidal amplitude (normalized RMS variability (NRMS)) as a metric for stationarity, low-mode M2 internal tides with NRMS< 0.5 are stationary over 25% of the deep ocean, particularly near the generation regions. The M2 RMS variability tends to increase with increasing mean amplitude. However, the M2 NRMS variability tends to decrease with increasing mean amplitude, and regions with strong low-mode internal tides are more stationary. The internal tide beams radiating away from generation regions become less stationary with distance. Similar results are obtained for other tidal constituents with the overall stationarity of the constituent decreasing as the energy in the constituent decreases. Seasonal variations dominate the RMS variability in the Arabian Sea and near-equatorial oceans. Regions of high eddy kinetic energy are regions of higher internal tide nonstationarity.