Re: BUFR variables with shape > 1

• To: Kevin Baggett <kevinb@xxxxxxxxxxxxx>
• Subject: Re: BUFR variables with shape > 1
• From: Robb Kambic <rkambic@xxxxxxxxxxxxxxxx>
• Date: Thu, 10 Aug 2006 15:50:22 -0600 (MDT)

On Tue, 8 Aug 2006, Kevin Baggett wrote:

Robb,
I am using some of your classes (BufrDataExtractor in particular) to get out 
some data for files I have. Some of the data has a shape of, say, 2, but when I 
use the getFloatData, it returns the first dimension all at once and then the 
second dimension. When I use the NetCDF GUI, it prints out alternating values 
from each dimension. Where in the code is this alternating value approach 
accomplished?

Kevin,

interpreting  compressed data is a much more difficult task then uncompressed.  
compressed data is a different animal, all the values for the fields are 
grouped together. to get one obs, you have to read the values using astride of 
numObs. the code is in BufrIosp.java, readDataCompressed, line 532

                 } else if( bd.isNumeric() ) {
                     ii = (IndexIterator)iiHash.get( v.getShortName() );
                     float[] bufrdata = bd.getFloatData();
                     //System.out.println( "bufrdata.length ="+
bufrdata.length
                     //+" = "+ numObs );
                     for( int m = 0; m < numObs && m != end; m++ ) {
                        for( int n = m; n < bufrdata.length; n += numObs
) {
                           ii.setFloatNext( bufrdata[n] );
                        }
                     }



the global attribute in the index; compressdata = true determines how the read 
will happen.

robb...

==============================================================================
Robb Kambic                                Unidata Program Center
Software Engineer III                      Univ. Corp for Atmospheric Research
rkambic@xxxxxxxxxxxxxxxx                   WWW: http://www.unidata.ucar.edu/
==============================================================================