Examples:  NetCDF & MatLAB
==========================

These directories contain examples of MatLAB commands for reading,
creating, and modifying NetCDF files.  All the scripts use Chuck
Denham's MatLAB package which can be obtained at:

  "MexCDF and NetCDF Toolbox For Matlab-5&6"
  http://woodshole.er.usgs.gov/staffpages/cdenham/public_html/MexCDF/nc4ml5.html

and must first be installed.


The NetCDF files generated here are intended for use with Ingrid:

  http://ingrid.mit.edu/

As a result, they approximately follow the simple COARDS convention
which is described at:

  http://www.unidata.ucar.edu/packages/netcdf/conventions.html
  http://ferret.wrc.noaa.gov/noaa_coop/coop_cdf_profile.html


The key items for the use of NetCDF files within Ingrid are:

  1) Although coordinate names such as "lat", "lon", "latitude",
     etc. will work without any problems for plotting data, it seems
     that the names "X" and "Y" sould be preferred for the coordinate
     names so that calculations (eg. curl) can be applied to those
     data sets.  For an example see:

       TRENBERTH_tau_mask/EH3_notes_mask.txt

     where both naming styles are used but the latter "X", "Y" naming
     scheme should be preferred.

  2) In order to work with the Ingrid "plot land" option, dimensions
     should have units of "degree_east" and "degree_north" as, for
     example, specified by:

       nc{'X'}.units = 'degree_east';
       nc{'Y'}.units = 'degree_north';

     The actual dimension names (eg. "lat" or "Y") are unimportant.

  3) It is generally a good idea to set the "gridtype" to 1 or 0 to
     indicate that the coordinate "wraps" or not.  The gridtype
     attribute allows Ingrid to, for instance, plot longitudinal data
     over "wrapped" ranges such as -100 to 500deg.  The attribute can
     be set using commands such as:

       nc{'X'}.gridtype = ncint(1);
       nc{'Y'}.gridtype = ncint(0);

  4) Ingrid does seem to correctly handle the "_FillValue" attribute
     and it can be set using a MatLAB command such as:

       nc{ idu }.FillValue_ = ncdouble(-9999.0);

     [ Note how the "_" character is moved to the end of the attribute
       name since MatLAB cannot correctly handle variable names with a
       leading underscore. ]