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

netCDF questions...

This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.

  • Subject: netCDF questions...
  • Date: Wed, 22 Jun 1994 11:08:54 -0600
>From: address@hidden (Rob -Not from Bloom County- Binkley, EIT)
>Organization: NASA Dryden FRC
>Keywords: 199406180327.AA22869 netCDF

Rob,

>I have installed the netCDF package OK.  I also printed out the user's guide 
>2.3.  Are there any simple example programs around?  Something simple to get 
>started ...

The simplest way to get started is to use the ncgen and ncdump routines that
are a part of the netCDF package.  ncgen can be used to create the skeleton
of a C or Fortran routine for data that you would like to store (i.e. it
would create a source file with all of the pretinent calls to the netCDF
interface library).  I suggest that you read the section concerning ncgen
in the User's Guide for more specific explanation.

>I am recording a lot of sensor data from an aircraft.  The data looks like:
>    A frame of 8000 parameters (mixed floats, ints, shorts, and logicals).
>Each frame has frame received time associated with it.

OK.

>I have looked at the user's guide.  I believe I want to define "records" to 
>be written to the netCDF file.  Do I define this using a collection of 
>"vectors"?  Time would be the UNLIMITED dimension...

You could accomplish your objective in a number of ways.  The slowest would
be to write each variable to the file for the current "record" (here I
am assuming that time would be your record/unlimited dimension).  The fastest,
and least netCDF-like would be to write vectors of floats, vectors of ints,
etc. for each record.  The problem with this is that the generality of the
netCDF (i.e. the ability to save meaningful meta data with the data itself)
is lost (or, at least, made very difficult).  The whole purpose of the
netCDF is to have a self-describing data form that generic software can
use (by generic, I mean software that is NOT written with specific knowledge
of the data in the file).  This notion automatically makes a tradeoff between
speed and generality.  It may be the case that your need to write a huge
number of parameters to a file in real time would preclude you from using 
the netCDF.  To me this doesn't mean that the netCDF might not be the best
form for data that you will want to access regularly later, but, rather, that
it might not be the best form for "ingesting" the data.

>I believe I'm missing a concept here...variables and dimensions.

This notion is the common one concerning multi-dimensional data objects.  The
one constraint in the netCDF is the ability to have only one unlimited 
dimension.

Tom Yoksas