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

Re: Vertical dimension in NJ



Hi Robert:

The problem is that the vertical coordinate is not being recognized. Section 4.3 of CF 
specifies a vertical coordinate is identified by units of pressure or the 
"positive" attribute.

The older version of the code was making anything not recognized into a 
vertical coord, but that was causing errors in other files, so i had to change 
to conform to the spec.

Can you let the writer of the file know, hopefully they can fix at the source.
Sorry for the trouble,

John

Robert B. Schmunk wrote:

Take a look in

ftp://ftp.giss.nasa.gov/outgoing/caron/

There are two files.

  tavg.00005.01.01.nc is the original.

  tavg.00005.01.01x.nc is the one I "fixed" by adding
     the positive attribute

There are about 8 variables which have a vertical
dimension. NJ 2.2.18 recognizes them in the original
file, but NJ 2.2.20 only in the "fixed" file.

By "recognize", I mean that the hasVerticalAxis() method
in the variables' CoordinateSystems is true.

rbs







On Jun 27, 2007, at 18:45, John Caron wrote:

im not sure whats going on, can yo send me the file?

Robert B. Schmunk wrote:
On Jun 27, 2007, at 17:43, John Caron wrote:
Do you know what Conventions are being used (ie what CoordSysBuilder class is being used?)
If I add some diagnostics right after opening the dataset
    NetcdfFile ncf = NetcdfDataset.acquireFile (url.toString ( ), null);
    NetcdfDataset ncd = new NetcdfDataset (ncf);
        CoordSysBuilder csb = new CoordSysBuilder ( );
           csb.buildCoordinateSystems (ncd);
System.out.println ("csb uses convention " + csb.getConventionUsed ( ));
I get that
  csb uses convention _Coordinates
Same result using either NJ 2.2.18 or 2.2.20.
rbs

Robert B. Schmunk wrote:
John,
I'd like to check with you about a change that seems to have occurred
between NJ 2.2.18 and 2.2.20 when it comes to recognition of vertical
dimensions.
After a query from a user, I was tracking down a change in behavior
between two versions of my Panoply application and found that it resulted from updating from using NJ 2.2.18 to NJ 2.2.20. In his dataset (tagged
with the convention attribute value "CF-1.0", there is a dimension
described as
   double zt(zt=19);
     :axis = "Z";
     :edges = "zt_edges";
     :long_name = "depth of the t grid";
     :standard_name = "depth";
     :units = "m";
If Panoply uses NJ 2.2.18, this dimension is recognized.
If Panoply uses NJ 2.2.20, this dimension is not recognized, but it
can be made recognizable by adding a
     :positive = "down";
attribute.
At this point my main question is whether this change in behavior
was deliberately made? If so, what was the reasoning for making it?
Thanks for your help,
rbs
--
Robert B. Schmunk, address@hidden
NASA Goddard Institute for Space Studies, 2880 Broadway, New York, NY 10025

--
Robert B. Schmunk, address@hidden
NASA Goddard Institute for Space Studies, 2880 Broadway, New York, NY 10025


--
Robert B. Schmunk, address@hidden
NASA Goddard Institute for Space Studies, 2880 Broadway, New York, NY 10025