Lateral Dynamics

The lateral dynamics are controlled in the namelist.dynamics. We have decided to experiment with the lateral eddy viscosity rn_ahm_0_lap, the biharmonic operator ln_dynldf_bilap, and the lateral boundary conditions for momentum set in the namelist.lateral.

Lateral Eddy Viscosity

The lateral eddy viscosity controls how dissipative the simulation is. Since we are forcing tides at the Juan de Fuca boundary we would like the visocisty to be as small as possible in order to prevent tidal phase lags within the Strait of Georgia. We have had some success at reducing the lateral viscosity as outlined in the table below.

Simulation	rn_ahm_0_lap	ln_apr_dyn	rn_avevd	Notes
\(\nu=100\)	100	true or false	10	stable
\(\nu=60\)	60	false	10	stable
\(\nu=60\)	60	true	10	unstable near islands
\(\nu=60\)	60	true	100	stable
\(\nu=50\)	50	false	10	unstable
\(\nu=40\)	40	true	60	stable
\(\nu=40\)	40	true	50	unstable near islands

These simulations were run on salish with model time seven days. The boolean ln_apr_dyn in namelist.surface controls forcing at the free surface due to atmospheric pressure. We are having difficulty with stability when this is set to true and the viscosity is lower than 100. Unfortunately, this type of forcing is likely important when modelling storm surges.

The parameter rn_avevd in namelist.dynamics controls the vertical eddy viscosity only in areas where/when the stratification is statically unstable. When the model is presented with an unstable stratification it locally increases the amount of vertical diffusion which effectively mixes the unstable region (see Vertical Mixing). Through our analysis, we have observed that the stability of the model at lower viscosities depends on how it treats vertical mixing. Thus, this parameter is likely important for achieving stability at low lateral viscosity. This must be investigated further.

We are still having trouble with stability during our spin up runs. See Spin-up Runs.

Biharmonic Operator

The biharmonic operator dissipates energy selectively at smaller scales. It is a fourth order diffusive operator in the momentum equations and is chosen by modifiying ln_dynldf_bilap in namelist.dynamics. It can be used in conjuction with the second order laplacian operator. The AMM configuration employs the bilaplacian with rn_ahm_0_blp=-1e10 and \(\nu=60\) for the laplacian operator. Note that AMM also uses free slip lateral boundary conditions and s-coordinates.

The biharmonic operator can be used in conjunction with the second order laplacian operator. Under the current resolution, typical values for the operator coefficient should be around rn_ahm_0_blp=-2000. Decreasing the magnitude of this parameter has some stablizing effect with little change in the maximum currents. However, the simulations with this operator in use still display overturning and poor behaviour in the vertical salinity profiles. A summary of simulations is given below.

Simulation	rn_ahm_0_lap	rn_ahm_0_blp	rn_avevd	Notes
apr60_nu50	50	none	60	unstable near islands
apr60_nu50_bi2000	50	-2000	60	unstable near islands
apr60_nu50_bi4000	50	-4000	60	unstable near islands
apr60_nu50_bi1000	50	-1000	60	stable, poorly behaved salinity

Lateral Boundary Conditions

Currently we are using partial slip boundary conditions with rn_shlat =0.5 in namelist.lateral. No slip conditions are applied when rn_shlat =2 and free slip when rn_shlat =0.

At \(\nu=50\), we have seen some stablizing features as we take the lateral boundary torwards no slip. It seems that the no slip conditions change the location of the maximum velocities in the island regions, which can have an affect on the mixing. Our concern with no slip stems from resolving the boundary layer. We fear that using no slip BCs will leave the boundary layer unresolved, especially at lower viscosity.

Simulation	rn_ahm_0_lap	rn_shlat	rn_avevd	Notes
partial25/apr100_nu50	50	0.25	100	unstable at Stuart Island
apr60_nu50	50	0.5	60	unstable near islands
partial1/apr60_nu50	50	1	60	stable
noslip/apr60_nu50	50	2	60	stable