[MITgcm-support] [Fwd: Auto-discard notification]

[Jean-Michel Campin]

Hi Eli,

> > Hello! I have downloaded the MITgcm code and using it to
> > investigate the red sea. To this end I'm running the model
> > (version 56) with exf, cal, obcs packages (just one open
> > boundary on the east side). The red sea is a narrow(~300 km)
> > and elongated basin (~2000 km). As the main axis points to
> > the NW, about three quarter of the grid points are 'land'
> > points (indicated by zero at the bathymetry input file). At
> > one of the test experiment I conducted, I have found that
> > computation time for land points is almost the same as the
> > 'sea' grid point (is it correct?). Following this, I would
> > like to rotate the main axis by ~45 degree which will allow
> > me to reduce the number of grid points dramatically and my
> > computation time too. So my question is if there is any
> > smart way to do this, but still preserve real Coriolis
> > parameters (is there a way to insert as an input file ?)
> > and volume conservation. Alternatively is there a way to
> > reduce somehow computation time of 'land grid point'
> > (which will make the work much easier). thanks Eli

The simplest way to go would be to rotate by 45^o and
to change the coriolis initialisation: model/src/ini_cori.F
I also added recently a way to read coriolis (fCori, fCoriG
& fCoriCos) from files (see at the end of ini_cori.F) but
you still need to change the default to activate this part
of the code.
Are you using cartesianiGrid or sphericalPolarGrid ? 

Jean-Michel