Please note that while the MathWorks now supports netCDF, it does not
support the NetCDF Toolbox, so please do not ask them any questions
about it.  For that matter, I do not support it either since I did
not write it.  If you use it, you are completely on your own.


SETTING YOUR PATH ======================================================

First, a word of warning.  If you use the NetCDF Toolbox, you will not 
be able to easily access the help for MATLAB's native netcdf package in 
R2008b and later.  "help netcdf" will always go to the NetCDF Toolbox,
even if it comes last on your path.  An annoying workaround is to use
"help netcdf/Contents", but a better solution is to rely on doc, i.e.
"doc netcdf".

In order to use the netcdf toolbox, you should add the following 
paths to your matlab path.

addpath /path/to/wherever/you/unpacked/this/netcdf_toolbox/netcdf 
addpath /path/to/wherever/you/unpacked/this/netcdf_toolbox/netcdf/nctype
addpath /path/to/wherever/you/unpacked/this/netcdf_toolbox/netcdf/ncutility

or if on Windows, 

addpath C:\path\to\wherever\you\installed\this\netcdf 
addpath C:\gpath\to\wherever\you\installed\this\netcdf\nctype
addpath C:\path\to\wherever\you\installed\this\netcdf\ncutility

It could reside on another drive as well.

And remember to remove any previous version of the netcdf toolbox from
your path as well.



Windows Users Take Note ===================================================

If you are using R2008a or below, then read the next few blurbs.

The "netcdf.dll" should be somewhere on your windows system path, BUT
NOT on your matlab path.  If you type

    >> which ( 'netcdf', '-all' );

from the matlab command line, you should only see one m-file, and you
shouldn't see the DLL at atll.

If you are using R2008b (or peeking into the future, R2009a or higher), 
then don't bother.  You don't have to worry about the netcdf.dll anymore.  
Woo-hoo!



AUTOSCALING ===============================================================

If you wish to automatically scale your data and have fill values 
replaced by NaNs, you should define a global structure called 
"nctbx_options" as follows:

    >> global nctbx_options;
    >> nctbx_options.theAutoscale = 1;
    >> nctbx_options.theAutoNaN = 1;

You can place this code in your startup.m if you don't want to 
continually have to do this upon starting matlab.