Using Windows-based GrADS to examine TRMM orbit data files

[Arlindo da Silva]

On Mon, Jul 21, 2008 at 8:11 AM, Mr Aaron Pratt <aspratt2000 at yahoo.com> wrote:
> Hello,
>
> I am trying to create plots derived from the TRMM 2A12 orbital product (it's
> the hydrometeor profile data) . It's not gridded data, but rather in orbital
> swaths.

Well, in a way swath data is gridded, just the I/J axis do not
correspond to lon/lat...

The files are HDF files, so that's why I thought that GrADS might be>
> able to open them. However, I wasn't sure if the HDF files needed to be
> converted into a GrADS type file (like a .ctl file and the corresponding
> .dat file). I was told that the files could be converted through the
> following link: http://disc2.nascom.nasa.gov/data/TRMM/; however I don't see
> anywhere on the website where this is possible. Thanks in advance for any
> help that can be provided.
>

Reading HDF-EOS swatch data is a common need, but not something that
can be directly done in GrADS at this point.  The bottom line is, the
data needs to be remapped to a regular lat/lon grid (I stress *remap*
because linear interpolation is not something that should be applied
to precipitation data). This can be accomplished in2 ways,

1) Implicitly by GrADS if you create a proper PDEF file. Nearest
neighbor interpolation is the safest, but mass conserving schemes
could be implemented as well. I believe a generic utility could be
written to automatically create PDEF files for swath data, or at at
least for the HDF-EOS variety.  One could also write a GrADS extension
along the lines of the g2() extension for doing the same thing on the
fly.

2) Explicitly, by writing a file conversion program  that does the
remapping and writes out, say, a grads binary file. You can also
convert the swath data to station data. Such utilities may already
exist.

For a quick inspection of the data you can always "cheat.".  Swath
data is after all given on a rectangular array. You can pretend the
crosstrack axis is longitude and along track is latitude, and write a
ctl/ddf that lies to GrADS. You will be able to look at your numbers,
but the map overlay will not make any sense. As I said, quick and
dirty.

Brian, Jennifer: what are your plans for swath data? Do you see it as
gridded or station data? Both representations can be quite useful,
depending on the nature of the dataset.

     Arlindo

--
Arlindo da Silva
dasilva at alum.mit.edu