About Nesting in ROMS

[cecilia]

Someone knows if:

1) Is there a note describing how to do nesting in ROMS ? or someone could explain to me how I could work with mother and nested grids using ROMS ?


2) Is there a software (may be matlab tools) to prepare mother and nested grids ? and where could I find it ?

Thanks

[kate]

1) Is there a note describing how to do nesting in ROMS ? or someone could explain to me how I could work with mother and nested grids using ROMS ?

2) Is there a software (may be matlab tools) to prepare mother and nested grids ? and where could I find it ?

Thanks

The Rutgers ROMS does not yet do nesting all in one go. We run the big grid, then the next grid using boundary conditions from the parent, and so on. There is work underway to be able to run the nests simultaneously.

The AGRIF ROMS does do nesting and they have the matlab tools already. The link was posted in one of these recent threads...

[cecilia]

Kate,

Then, I should implement and run ROMS 3 times if I want to use one parent grid and two nested grids ?

or

Is it possible to run the model in a sequential fashion after defining the three grids ? If this is the case how I should implement ROMS (grid parameters format) and build the NetCDF grid files if I am using Seagrid ?

[kate]

Yes, that's right. Save the output of the parent grid often enough to use as boundary conditions for the children. We've been using 1-5 days. We've got a variety of matlab scripts for extracting the boundary condition file. John Wilkin has been using the station data, setting up stations all along the children's boundary positions in the parent grid, while Enrique and I work from the averages files.

I personally don't mind separating the grids like this. If the outer domain is going to be at all troublesome, solve its issues before adding the overhead of the children.

Edit: just saw the bottom part of your post. Build each grid independently, making sure it suits your needs on its own. The nests can be 3 times finer on the same orientation or off at an angle counting on the interpolation scheme, just make sure it is within the outer domain.