HI Christian,
Very interesting... it would be great to see if this mechanism could be
used to generate an FMRC aggregation (i.e. aggregating along a time
dimension) simply by giving it a directory of files, with no need to
(manually) create NcML.
Cheers, Jon
From: netcdf-java-bounces@xxxxxxxxxxxxxxxx
[mailto:netcdf-java-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Christian D
Ward-Garrison
Sent: 22 March 2010 14:47
To: Christian D Ward-Garrison
Cc: netcdf-java@xxxxxxxxxxxxxxxx
Subject: Re: [netcdf-java] Aggregation without NcML
Well, it's pretty clear that John didn't intend for the aggregation
machinery to be used directly (it's not in the public Javadoc), but
after a couple hours with the debugger, I think I've hacked together a
fairly clean solution:
package gov.usgs.shodan;
import java.io.File;
import java.io.IOException;
import ucar.nc2.NetcdfFile;
import ucar.nc2.dataset.NetcdfDataset;
import ucar.nc2.ncml.Aggregation;
import ucar.nc2.ncml.AggregationUnion;
import ucar.nc2.util.CancelTask;
/**
*
* @author cwardgar
*/
public class Mlar {
public static void main(String[] args) throws IOException {
File cldcFile = new File("C:/Documents and
Settings/cwardgar/Desktop/cldc.mean.nc");
File lflxFile = new File("C:/Documents and
Settings/cwardgar/Desktop/lflx.mean.nc");
NetcdfFile cldcNcFile =
NetcdfFile.open(cldcFile.getAbsolutePath());
NetcdfFile lflxNcFile =
NetcdfFile.open(lflxFile.getAbsolutePath());
NetcdfDataset unionDataset = unionNetcdf(cldcNcFile,
lflxNcFile);
System.out.println(unionDataset);
unionDataset.close(); // Closes cldcNcFile and lflxNcFile.
}
public static NetcdfDataset unionNetcdf(NetcdfFile ncFile1,
NetcdfFile ncFile2) throws IOException {
NetcdfDataset aggDataset = new NetcdfDataset();
AggregationUnionOpenDataset union = new
AggregationUnionOpenDataset(aggDataset);
union.addOpenFile(ncFile1);
union.addOpenFile(ncFile2);
union.finish(null);
aggDataset.finish();
return aggDataset;
}
// Aggregation.Dataset is a non-static nested class. So, to subclass
it, I must create an enclosing
// Aggregation subclass.
public static class AggregationUnionOpenDataset extends
AggregationUnion {
public AggregationUnionOpenDataset(NetcdfDataset unionDataset) {
super(unionDataset, null, null);
}
public void addOpenFile(NetcdfFile openFile) {
addDataset(new OpenUnionDataset(openFile));
}
public class OpenUnionDataset extends Aggregation.Dataset {
private NetcdfFile openFile;
public OpenUnionDataset(NetcdfFile openFile) {
super(openFile.getLocation());
this.openFile = openFile;
}
@Override
public NetcdfFile acquireFile(CancelTask cancelTask) throws
IOException {
return openFile;
}
}
}
}
Note that this example uses the same datasets that the NcML tutorial
(http://www.unidata.ucar.edu/software/netcdf/ncml/v2.2/Aggregation.html)
does and will print a similar CDL.
Now I've just got to figure out how to aggregate on an existing
dimension and scan a directory.
-----netcdf-java-bounces@xxxxxxxxxxxxxxxx wrote: -----
To: "netcdf-java@xxxxxxxxxxxxxxxx" <netcdf-java@xxxxxxxxxxxxxxxx>
From: Christian D Ward-Garrison <cwardgar@xxxxxxxx>
Sent by: netcdf-java-bounces@xxxxxxxxxxxxxxxx
Date: 03/19/2010 07:30AM
Subject: [netcdf-java] Aggregation without NcML
Hello,
Is it possible to aggregate NetCDF datasets from within my
application--that is, without creating an NcML document for them
beforehand?
Suppose I want to write a function that takes 2 arbitrary NetcdfFile
objects and returns another NetcdfFile object that is the union of them.
What's the best way to do this?
Thanks,
Christian Ward-Garrison
_______________________________________________
netcdf-java mailing list
netcdf-java@xxxxxxxxxxxxxxxx
For list information or to unsubscribe, visit:
http://www.unidata.ucar.edu/mailing_lists/
