[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[IDV #OCI-470088]: Plotting WRF data with IDV

Hi Gary-

> Well, in my lab this summer we've (finally) been striving to use IDV,
> specifically to view WRF model output.  It is great that IDV just sees
> the wrfout files right away!  However we have run into a brick wall when
> it comes to performance, but it could also be that we're just not using
> it correctly.  Plotting isosurfaces for grid files with dimension
> greater than about 200x200 seems nearly impossible, as does plotting any
> 3-D grid.  We are running IDV on linux boxes with at least 1Gb of RAM,
> and we did increase the RAM allocation to 1024 in the runIDV script
> following an earlier suggestion by Jeff.

If you only have 1 Gb on your system, I would not set the value to
be more than 700 otherwise, you'll encounter disk swapping which
will really degrade your performance, particularly on windows.
 
> We are working with arrays that are typically 400x400x31 elements.  In
> addition to the runIDV mod, Jeff had also suggested editing the
> "properties", and increasing the stride to every Xth point... the
> problem with that is when you do this for a 3-D grid variable, it gives
> an error and says that it cannot find or plot the data - unrecognized
> vertical coordinate.  The stride subsetting seems to work fine for 2-D
> variables.  However, the 2-D variables are easy to plot in the first
> place, so it is really the 3-D varieties where we most need the
> subsetting feature to work.

I thought this was fixed with 1.3b2, but apparently not.  If you
could provide us with a sample file that would be great.

The problem is dealing with the staggered grids.  For staggered
grids, we end up creating two domains - the staggered and the
interpolated unstaggered grid which doubles the memory useage.
It also means that for any computation (like creating wind from
U & V) the grids need to be transformed from grid coordinates
to a common interpolation grid.  We do not yet have an elegant
solution for this.  In the end, it seems like people want to
plot the data on unstaggered, pressure grids and we are encouraging
people to use the post process WRF steps to create the necessary
GRIB files.

> It could be that there is confusion witht he WRF vertical coordinate in
> that circumstance?   We are using the 1.3b2 version of IDV.  Any
> suggestions would be appreciated.  If needed, we would be happy to
> supply a wrfout file of the type with which we are struggling if it
> would help.

When the reduction is done, something is not being calculated
correctly.  If you give us a file, we'll track that down.

Thanks.

Don

Ticket Details
===================
Ticket ID: OCI-470088
Department: Support IDV
Priority: Normal
Status: Open