I am developing a prototype to save SAR data at various levels and processing
stages in HDF5 format.
In order to properly describe this data (e.g. the side looking geometry or
orbit information in form of state vectors), I would need to generate a
larger number of attributes that would benefit from a structure in groups.
The majority of the attributes would apply to all data sets. Sometimes, you
will have several bands in a data set with a very limited number of
attributes specific to the band (e.g. its polarization). Is there a standard
way of going about saving this type of metadata structure in HDF5?
Another somewhat related question is how to handle map projection
information for geocoded data. Any best practice for that?
···
--
View this message in context: http://hdf-forum.184993.n3.nabble.com/satellite-metadata-tp1515627p1515627.html
Sent from the hdf-forum mailing list archive at Nabble.com.
Please, take a look at the CF metadata conventions: http://cf-pcmdi.llnl.gov/
It assumes that you use netCDF, but using netCDF-4 you can write well structured HDF5 files, or you can follow the conventions using standard HDF5 functions.
I have participated in a small project called ADAGUC where we tried to address also the meta data for remote sensing datasets: http://adaguc.knmi.nl/
Good luck, Richard
···
On 09/17/2010 02:52 AM, rudigens wrote:
I am developing a prototype to save SAR data at various levels and processing
stages in HDF5 format.
In order to properly describe this data (e.g. the side looking geometry or
orbit information in form of state vectors), I would need to generate a
larger number of attributes that would benefit from a structure in groups.
The majority of the attributes would apply to all data sets. Sometimes, you
will have several bands in a data set with a very limited number of
attributes specific to the band (e.g. its polarization). Is there a standard
way of going about saving this type of metadata structure in HDF5?
Another somewhat related question is how to handle map projection
information for geocoded data. Any best practice for that?
Richard,
Richard van Hees wrote:
Please, take a look at the CF metadata conventions:
http://cf-pcmdi.llnl.gov/
It assumes that you use netCDF, but using netCDF-4 you can write well
structured HDF5 files, or you can follow the conventions using standard
HDF5 functions.
I have participated in a small project called ADAGUC where we tried to
address also the meta data for remote sensing datasets:
http://adaguc.knmi.nl/
The other data format addressed in the prototype is actually netCDF. I have
followed the CF conventions and I am working with some netCDF folks to check
the correctness of the implementation. The "extra" metadata beyond the CF
convention is one of the remaining questions for the netCDF prototype as
well. There are a number of ways to group them. I am searching for a
solution that looks somewhat intuitive to the netCDF/HDF5 user communities.
Some of the "common practice" with the netCDF format was a little puzzling
for me. The projection is usually defined according to the CF convention
with individual projection parameters, with the ArcGIS projection file type
string as well as the proj4 definition of the projection. That sounds a
little overkill to me but then there are two additional layers for
latitude/longitude per pixel added on top of that. That got me into the use
of compression and netCDF-4 with HDF5 in the background. Otherwise the file
sizes cross the GB limit big time with images in the order of 10000x10000
pixels.
I was hoping that the HDF5 world has a little simpler convention. Thanks for
the link to ADAGUC. Very nice use case.
Cheers,
Rudi
···
--
View this message in context: http://hdf-forum.184993.n3.nabble.com/satellite-metadata-tp1515627p1517583.html
Sent from the hdf-forum mailing list archive at Nabble.com.
Thanks, I was not aware of its existence. I will definitely subscribe to this list.
Greetings, Richard
···
On 10/01/10 10:00, Martin Raspaud wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Richard van Hees skrev:
On 09/17/2010 02:52 AM, rudigens wrote:
I am developing a prototype to save SAR data at various levels and
processing
stages in HDF5 format.
In order to properly describe this data (e.g. the side looking
geometry or
orbit information in form of state vectors), I would need to generate a
larger number of attributes that would benefit from a structure in
groups.
The majority of the attributes would apply to all data sets.
Sometimes, you
will have several bands in a data set with a very limited number of
attributes specific to the band (e.g. its polarization). Is there a
standard
way of going about saving this type of metadata structure in HDF5?
Another somewhat related question is how to handle map projection
information for geocoded data. Any best practice for that?
Please, take a look at the CF metadata conventions:
http://cf-pcmdi.llnl.gov/
It assumes that you use netCDF, but using netCDF-4 you can write well
structured HDF5 files, or you can follow the conventions using standard
HDF5 functions.
I have participated in a small project called ADAGUC where we tried to
address also the meta data for remote sensing datasets:
http://adaguc.knmi.nl/
Good luck, Richard
The cf conventions are being actively developed at the moment for satellite
data. Take a look here:
http://mailman.unidata.ucar.edu/mailman/listinfo/cf-satellite
Best regards,
Martin
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/
iQEcBAEBAgAGBQJMpZUTAAoJEBdvyODiyJI4T3QH/jvRy+uB0BRJuxK+r+P1NOkP
JjjSWWaPl1OTJTrbtPhgw0VIszyfaFgi0c9+xjltai2N5W5xJXTIRITFOZccw+RQ
F9QQebmjGwKdhm5aBx9MdaIR6FiLVsYQB0fvoPepeoB/TKEZoOziRUM2rqNZbOXa
/vDFYqlG27Leqf2i/mlU57t3Pl+V3jSW0WOoQuCkAMGWk74ujfiIuo07wWtivI5b
d7oH/GRNFMl74BSbyrdB9MRRVtO28fmerRM5/mUCfE5FzSqnRf9rCWbxpxscik7H
AnN1hLFdjlRfMCwDsPWH9/dDFo+9Djp7ycv2HDtHc4ofzPJDkEpPhAYLPQLogLE=
=zcIb
-----END PGP SIGNATURE-----
_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@hdfgroup.org
http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org
