The codes and sample data can be found at ftp://sprite.llnl.gov/pub/fiorino/grads/data. See the README file: ftp://sprite.llnl.gov/pub/fiorino/grads/data/grads/README for more information on the data in ftp://sprite.llnl.gov/pub/fiorino/grads/data/grads/grads.test.data.tar (it is 4.6 Mb).
Working With Station Data In GrADS
Mike Fiorino
Program for Climate Model Diagnosis and Intercomparison
Lawrence Livermore National Laboratory
Livermore, CA
fiorino@llnl.gov
19 October 1995
while visiting
Meteorological Research Institute
Tsukuba, Japan
1. Introduction
The direct access format (straight binary) of station data in
GrADS has inhibited some fortran programmers from converting
their data for use in GrADS. The new version of GrADS (1.5.1.17)
I developed while visiting MRI now supports the "sequential"
option. That is, GrADS now works with "f77" binary station data
in the same way as gridded data.
The original direct access (C code) format is still supported and
is recommended for two reasons: 1) there is no true standard for
unformatted binary output in f77; and 2) the data files will be
larger. However, on most unix platforms, station data written in
fortran should work and the hardware options (i.e., big_endian
and little_endian) can also be specified so that, for example,
station data created on a little endian machine (e.g., linux) and
be displayed on big endian machine (e.g., sgi).
2. Sample Data and fortran code.
There are three basic converion problems: 1) single level data
(i.e., "surface" obs); 2) upper air data WITHOUT sfc obs (e.g., a
rawindsonde sounding); and 3) upper air data WITH associated sfc
obs (e.g., a rawindsonde sounding with observations at the sfc).
The files have the naming convention:
*.txt - the original data in ascii
*.ctl - the GrADS .ctl
*2g.f - the f77 code to do the conversion
*.smp - the GrADS station map file created by the stnmap utility
*.obs - the binary data file
* = s sfc obs
= ua upper air only
= uas upper air + sfc obs
The f77 code and .ctl files contains documentation on the data
and code. The only thing I did not add to the .ctl which I
STRONGLY RECCOMEND YOU ADD is the type of the hardware the data
were created on in the options card. For example, if you ran the
code on a dec machine change
options sequential
to,
options sequential little_endian
and you'll be able to work with the data on any platform except
the cray.
3. Bugs --- the final bug is always the penultimate one
I have only tested the GrADS code and converion code only on the
dec alpha so I'm not sure how it will work on the other
platforms. While I expect no real trouble, please send me email
WITH the .ctl, code and sample data should you have problems.