Hi,
I am trying to figure out which lateral boundary conditions would be appropriate for my application, and I cannot seem to be able to find a complete description of the lbcs available in ROMS.
My first question is: which ones require an input file and which ones are fine by themselves?
I am trying to set up a channel with a flow that comes in at the top and leaves at the bottom. Which lbcs would you recommend?
I want a constant forcing at the upstream boundary for the entire length of the simulation. It seems that the "clamped" bc is the most adequate. It obviously requires an input value, so I created a netcdf file that provides my tracers distribution at the upstream boundary. The variables have the dimensions (# of timesteps, depth, y). Do I need to specify my clamped configuration for each time steps, or do ROMS will know that he will need to use this configuration for each time steps? In other words, should my variables have dimensions (# of timesteps, depth, y) or (1,depth,y) ?
As for my downstream boundary, I want the flow to be able to get out of the domain. For the info I could gather, it seems that a "radiation" bc is best. But I could not really find information on how this configuration works. Does it require an input file? if yes, what kind of input? Because what I want is just the flow to go out, whatever is at this boundary at that time.
The ROMS documentation is very brief when it comes to lbcs, so any information would be more than welcome!
Thank you,
Mat
Lateral boundary conditions (lbcs)
Re: Lateral boundary conditions (lbcs)
I just wrote an answer for a similar question here: viewtopic.php?t=2840#p10727
I've been implementing a similar setup viz., flow entering at 'top' (northern edge) and leaving at 'bottom' (southern edge). I'm using the RadNud option from the Marchesiello et al (2001) paper to impose the flow at the top. I think this is the 'gentler' way to do it than Clamping. It does require an input file that specifies what values (of T/S/u/v/ubar/vbar - depending on what you want to impose) you want to nudge to. I think your (1,depth,y) version of the boundary .nc file will work for all timesteps if you specify Clamped. It's what I use for my RadNud setup.
FYI, I have found that there tends to be high wavenumber noise near the southern boundary and a sponge layer helps kill it off. Hope this helps.
I've been implementing a similar setup viz., flow entering at 'top' (northern edge) and leaving at 'bottom' (southern edge). I'm using the RadNud option from the Marchesiello et al (2001) paper to impose the flow at the top. I think this is the 'gentler' way to do it than Clamping. It does require an input file that specifies what values (of T/S/u/v/ubar/vbar - depending on what you want to impose) you want to nudge to. I think your (1,depth,y) version of the boundary .nc file will work for all timesteps if you specify Clamped. It's what I use for my RadNud setup.
FYI, I have found that there tends to be high wavenumber noise near the southern boundary and a sponge layer helps kill it off. Hope this helps.