Fish() requires a *global* domain

Arlindo da Silva dasilva at OPENGRADS.ORG
Tue Feb 17 14:56:46 EST 2009

On Mon, Feb 16, 2009 at 10:08 PM, Isaac Tetteh <itetteh at> wrote:

> Hi Arlindo,
> Please I'm currently using Win 32 Superpack/Grads v1.9.0-rc 1.
> I'm using global NCEP/NCAR reanalysis data, which are at multiple pressure
> levels. These are available at
> .
> I have subset the data to my domain of interest eg -50 90 lon, -40 40 lat.
> Even for the moisture budget, I need to further narrow down my domain
> further to quantify the moisture in the area of interest to reflect the
> focus of my analysis.
> If you need further clarification, I will be happy to let you know you.

That is the problem: in order to compute streamfunction fish() uses a global
poisson solver; it does not work with anything that is not global. Although
you can implement poisson solvers for a local domain, in this case you have
to specify boundary conditions --- something that is not easy to properly
specify.   You can find in the literature variational algorithms for
computing streamfunction from (u,v) on a regional domain, but we do not have
this implemented yet. If you are incline to develop this, I'm sure it would
be very welcome by the community. I understand the algorithm, I just do not
have the time to code it right now. For your problem, though, it may be
simpler to compute chi/psi in the global domain, and then subset to your
domain of interest.


Arlindo da Silva
dasilva at
-------------- next part --------------
An HTML attachment was scrubbed...

More information about the gradsusr mailing list