5. Run tide only model - NOC-MSM/SEAsia GitHub Wiki

In a small domain the tides can be prescribed at the boundaries as depth invariant velocities and sea surface heights. For larger domains the spatially varying tidal potential is also used. These forcing are given as harmonic constants and are reconstructed on-the-fly, by NEMO, into timeseries. These initial stability tests with tides start from rest and so there is an unphysical transition period where the simulation spins-up. In regions where the tides are particularly energetic and there is a complex tidal propagation pattern to establish (particularly if there are narrow straits in the domain) the unrealistic spin-up phase can lead to model instability. The complex tidal pattern is the result of a final equilibrium between energy flux in at the boundary, propagation in the domain, and dissipation through bottom friction. During the spin-up unrealistically large velocities can occur. This can be alleviated by slowly ramping up the tidal forcing to reduced the shock to the system.

In \verb|EXP_tideonly|, we simulate 60 days of tidal evolution FIGURE. The spin-up of semi-diurnal constituents is fast (days). Longer period waves take longer \textcolor{red}{Jeff: ASK DAVE}. Plotting the magnitude of the maximum velocity, the sum of sea surface height squared helps visualise the evolution has detect problems as they arise. The 60 day test simulation highlights the 14 day spring neap cycle and its monthly modulation subject to the varying proximity of the moon to the Earth. Longer period modulations also exist according to the predictable celestial motions (typically each has diminishing influence on the total amplitude with increasing period). But rather than exhaustively check all the possible tidal phases a useful guide is to run a couple of months of tidal simulation to get an appreciation for the magnitude of the maximum tidal velocity in the domain and also to get an estimate for the maximum elevation due to the tides. With knowledge of the maximum expected tidal velocity the barotropic timestep can be appropriately lengthened to ensure stability or shortened to increase efficiency. With knowledge of the maximum tidal elevation, which typically happens at the coastal grid points, care can be taken to ensure the model grid cells do not dry out with the retreating tide (by prescribing the minimum water depth during the \verb|domain_cfg.nc| build process if necessary). Note that in the full simulation the addition of meteorological effects can act to enhance both the velocities and the surface elevation and so bathymetry and timestep modifications should not be too conservative to the theoretical limit.

60 day simulation from rest to verify model stability under tide only forcing EDIT FIGURE TO BE MAX OVER DOMAIN

In case there are troubles running with tides-only which you have reason to believe are independent of the boundary forcing, you should confirm that the simulation will timestep and maintain a state of rest when unforced. The boundaries can be deactivated with \verb|ln_bdy=.FALSE.|, (or setting them to their initial state: \verb|nn_dyn2d_dta=0| as a test of the boundary coordinates file, \verb|coordinates.bdy.nc|, file) and tides can be deactivate with \verb|ln_tide = .FALSE.| since they can also act as a body forcing away from boundaries if tidal potential forcing is used. An unforced simulation would provide a check that the NEMO and XIOS executables are properly built and that you have all the necessary ancillary files.

⚠️ **GitHub.com Fallback** ⚠️