i am trying to simulate the circulation of the North Aegean region. My problem is that after a year run zeta value ends to be approximately 5.7( !!!!) meters. I understand that probably one of the reasons (or the main ?) that causes the problem is that i am using the river forcing configuration (external forcing file and not ANA_PSOURCE flag) to simulate the Dardanelles flow into the north Aegean region and the fact that i am giving only the inflow and not the outflow from the Aegean sea. To be more precise, the flow follows the seasonal variability of the inflow to the north Aegean (inflow rate 622 km3/year, highest "discharge" end of July). I am asking if it possible to simulate a two layer flow using the river forcing option (the first 20 meters is the inflow from Dardanelles to the Aegean and from that depth and above is the outflow from the Aegean to Dardanelles) or in general if it possible to simulate a two layer flow for a case like this. I am using 651 svn revision, my grid spacing is 185x171x020, the boundary conditions for west and north edges is closed and for south and east is Chapman for zeta, Clamped for u,v,ubar,vbar,salt and temp and Gradient for tke. The grid stiffness ratios are: rx0 = 1.500031E-01 and rx1 = 4.846303E+00. I am attaching a figure showing zeta after the run is finished and my include/north_aegean.h file. Any suggestion or help will be highly appreciated.
The reason using the sources for a case like this is is challenging is that there is a global switch for imposing the tracer values on the source waters. In fact, you want to impose it if the flow is into the model and you want to not impose it if the flow is out of the model. That's why Marchesiello (2001) set up his boundary conditions the way he did, querying the model for the direction of phase speed and setting the nudging to exterior values point by point. Rather than clamped, we use a combination of radiation and nudging (RadNud).
i would like to ask a couple more things. From your answer and from Marchesiello's paper i understand that i have to set Radiation and Nudging boundary conditions for the tracers (salinity and temperature), for the 3D u,v momentum components and finally for the 2D u and v. These boundary conditions are both for south and east edges? The "main" and most important flow is from the east boundary (from Sea of Marmara to the north Aegean). My second and last question is this: do i have to modify the ana_hmixcoef.h and ana_nudgcoef.h files to set up the correct scales/factors for sponge layer and nudging?
You should look at those files very carefully. Sponging and nudging are two other questions - in our latest Northeast Pacific we do not sponge but we do nudge over a band around the edges. This is different nudging from the Marchesiello nudging, which is happening on the outside boundary points only. I personally do not want things done the way the distribution code has for ana_nudgcoef.h, which links the exterior nudging to the interior nudging timescale. It's a nudging to two different kinds of fields, in that we have fresh boundary conditions every 5 days while the climatology is monthly.
For LBC, I use RadNud only on the 3D fields. I use Chapman on the free surface, Flather on the 2D momentum.