>
>I am new to this list and an NCL novice. I plan to install NCL on a
>Linux machine running RedHat Enterprise v3.0. I would like to run NCL in
>batch mode to compute horizontal divergence of wind using uv2dv_cfd,
>which supposedly computes the divergence using centered finite
>differences. The program uv2dv_cfd supposedly requires that the input
>data be written to a latitude-longitude grid. The source files for which
>I want to calculate divergence contain numerical model data from the NAM
>(formerly the Eta), and eventually the WRF, which I can acquire/generate
>in either netCDF or GRIB format.
Welcome to the NCL world.
I think you will find NCL's netCDF and GRIB file handling
to be excellent.
There is no "supposedly" about it 'uv2dv_cfd' will compute
the divergence on a limited area lat/lon grid. However,
the requirement is that the grid be regular in the sense that
any grid location can be uniquely accessed vi *one* dimensional
lat/lon arrays. In netCDF parlance, these are called
coordinate variables.
>
>Question 1-- Assuming that I cannot use uv2dv_cfd on the native grids,
>what software do I use to write the (netCDF and/or GRIB formatted) data
>to a latitude-longitude grid, that can then be utilized by uv2dv_cfd?
>The software needs to be run from the command line (no GUI's please.)
>
The issue with the eta and WRF models is that the grids
require *two* dimensional lat/lon arrays. Twice, I have asked
the MMM/WRF people if they have a fortran or C code that
performs the calculation. There answer was "no".
If anyone has fortran or C code that does this we would
could make it built-in function.
Well NCL has interpolation functions that can interpolate scalars
to the lat/lon grid. However, u and v are vector components
and any proper interpolation would require that these
be interpolated as a pair.
Note: for visual purposes, perhaps even for relative vorticity,
using scalar interpolation (u and v separately) would probably
be ok. However, divergence is very sensitive to small differences
and I would be skeptical of the results.
>Question 2-- I also would like to compute horizontal gradients of a
>scalar. The data are written to HDF-EOS files, and I can also convert to
>netCDF.
No need, NCL can read HDF-EOS ... I speculate it is of type "grid"
>The data will include missing values, and is of limited area
>(not Global.) Anyone aware of how I could use NCL to compute the
>gradients (in batch mode)?
Again, the eta/WRF models require 2D lat/lon arrays which
uv2dv_cfd does not handle. [uv2dv_cfd does handle missing data.]
>Otherwise, what other software can I use?
I do not know. I would ask the WRF community. Maybe, you
will get a different answer.
Good luck
D
>Again, only command-line software.
>
>Any assistance will be greatly appreciated.
>
>Waylon
>
>--
>**********************************************************************
>* Waylon G. Collins Voice: (361)289-0959 *
>* NOAA/National Weather Service E-Mail: Waylon.Collins@noaa.gov *
>* 300 Pinson Drive *
>* Corpus Christi, TX 78406 *
>* USA *
>**********************************************************************
>
>_______________________________________________
>ncl-talk mailing list
>ncl-talk@ucar.edu
>http://mailman.ucar.edu/mailman/listinfo/ncl-talk
_______________________________________________
ncl-talk mailing list
ncl-talk@ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
This archive was generated by hypermail 2b29 : Wed Aug 17 2005 - 11:58:08 MDT